How to accelerate antimicrobial susceptibility testing

Clinical Microbiology and Infection 25 (2019) 1347e1355

Contents lists available at ScienceDirect

Clinical Microbiology and Infection journal homepage: www.clinicalmicrobiologyandinfection.com

Narrative review

How to accelerate antimicrobial susceptibility testing E.A. Idelevich*, K. Becker Institute of Medical Microbiology, University Hospital Münster, Münster, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 January 2019 Received in revised form 27 March 2019 Accepted 18 April 2019 Available online 2 May 2019

Background: Antimicrobial susceptibility testing (AST) results are crucial for timely administration of effective antimicrobial treatment, and, thus, should be made available to clinicians as fast as possible. In particular, increasing rates of multidrug-resistant organisms emphasize the need for rapid AST (rAST). Objectives: This article aims to provide microbiologists and clinicians with a critical overview of the current state of possibilities to accelerate AST. We also intend to discuss technical and strategic aspects of rAST, which may be helpful to academic researchers and assay developers in the industry. Sources: We have reviewed literature on rAST methods and their implementation in routine diagnostics. Content: Phenotypic rAST is universal, mechanism-independent and allows exact categorization, but it demands time for the microorganisms to start the growth and to express the response to antibiotics. Detection of selected resistance mechanisms is more rapid, but the interpretation of its clinical impact is limited. Technical challenges of phenotypic rAST include inoculum effect, delayed expression of resistance, lag phase and initial biomass increase in susceptible isolates. Criteria for a successful rAST assay are ease of use, random access, capacity for simultaneous testing of multiple specimens, affordability and financial attractiveness for industry. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS)-based AST seems to be particularly promising, as it can optimally be combined with MALDI-TOF MS identification. Direct testing from clinical specimens provides particularly early findings, with positive blood cultures being the most suitable specimen type. Polymicrobial samples and inoculum effect are serious obstacles for direct AST from other clinical specimens. Next to the technology improvement, optimization of pre-analytics and laboratory organization is essential. Implications: It appears feasible to generate an AST report within the same working shift; however, only affordable and easy-to-use rAST technologies have a chance to enter broad diagnostic routine. Efforts should be made by industry, authorities and academia to enable wide dissemination of rAST in clinical diagnostics. E.A. Idelevich, Clin Microbiol Infect 2019;25:1347 © 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

Editor: L. Leibovici Keywords: Antibiotic Direct susceptibility testing Inoculum effect Lag phase MALDI-TOF mass spectrometry Phenotypic susceptibility testing Rapid antimicrobial susceptibility testing Rapid diagnostics

Rapid antimicrobial susceptibility testing as cornerstone of modern infectious disease management The possibility to encounter an infection caused by a resistant pathogen increases with the currently rising rates of multidrugresistant organisms [1,2]. This increased risk of clinical treatment failure emphasizes the growing role of antimicrobial susceptibility testing (AST) in general and, in particular, approaches for rapid AST (rAST) [3]. The ultimate goal would be the availability of the pathogen's susceptibility profile at the time point of therapy

* Corresponding author. E.A. Idelevich, Institute of Medical Microbiology, University Hospital Münster, Domagkstr. 10, 48149, Münster, Germany. E-mail address: [email protected] (E.A. Idelevich).

initiation so that an appropriate antibiotic can be chosen from the very beginning [4]. Unfortunately, this is currently impossible and antimicrobial therapy for acute infections is still initiated as empiric treatment [5]. Nevertheless, antimicrobial treatment should be adjusted as soon as possible to target the causing pathogen [2,3,5]. The relevance of rAST for patient management and infection control [6e8] has been internationally recognized [2,3]. While plenty of rAST technologies have been suggested recently (Table 1), most of them are not available for clinical diagnostics and AST results in the routine practice are usually still available only on the next day [3]. Refraining from detailed explanation of specific technologies [9,10], we focus this review on critical discussion of technical and biological possibilities and limits of rAST, challenges and ways to overcome them, clinical needs for rAST and criteria for successful development of a rAST assay.

https://doi.org/10.1016/j.cmi.2019.04.025 1198-743X/© 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

1348

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

Table 1 Technologies for rapid phenotypic growth-based antimicrobial susceptibility testinga Technology

Short description

Reference

Disk-tube method

Growth-based method in liquid medium with visual evaluation of turbidity Bacterial growth is measured by using a medium containing a pH indicator (phenol red) Bacterial growth is detected by using a medium containing a redox indicator (resazurin) Visual evaluation of inhibition zones after abbreviated incubation Bacteria immobilized in the agarose matrix in a microfluidic channel; the growth of single cells is monitored using microscopy Optical growth detection in a liquid sample by the laser scattering method Optical growth detection by serial imaging in a liquid sample High-resolution mass measurement using microchannel cantilevers Growth detection by the measurement of heat production After an incubation in liquid medium, real-time PCR is used for quantification of DNA copies of either the 16SRNA genes or rpoB Luciferineluciferase assay produces light in the presence of ATP. The produced light is proportional to the bacterial ATP and, thus, to the microbial concentration The luciferase reporter phage is used to infect bacteria; quantifiable light is produced in case of bacterial growth Bacterial cells are immobilized on a surface, digital microscopy records microbial response to a single concentration of an antibiotic and software derives MIC values Bacterial colonization within a pillar-type grating is measured by phaseshift reflectometric interference spectroscopy in real time Assessment of drug-induced microbial lesions that lead to changes in morpho-functional parameters (e.g. membrane potential, cell size, amount of DNA) Incubation of microbial suspensions as microdroplets directly on MALDI targets, followed by a simple broth removal and MALDI-TOF MS measurement

Schneierson 1954 [51]

Colorimetric method utilizing a pH indicator Colorimetric method utilizing a redox indicator Disk diffusion method with short incubation Microfluidic agarose channel system with microscopic single cell growth tracking Forward laser light scattering Digital time-lapse microscopy Microbial cell mass measurement Isothermal microcalorimetry Real-time PCR ATP-bioluminescence

Luciferase reporter phage Morphokinetic cellular analysis

Intrinsic phase-shift spectroscopy on micropillar architectures Flow cytometry

MALDI-TOF MS direct-on-target microdroplet growth assay a

Rogers et al. 1955 [50] Sorensen 1959 [52] Barry et al. 1973 [56] Choi et al. 2013 [87] Idelevich et al. 2017 [17] Fredborg et al. 2013 [88] Godin et al. 2010 [89] von Ah et al. 2008 [90] Rolain et al. 2004 [38] Thore et al. 1977 [91]

Riska 1999 et al. [92] Descours et al. 2018 [93]

Leonard et al. 2017 [94] Pore 1990 [95]

Idelevich et al. 2018 [20]

The table is intended to present examples rather than exhaustive overview of technologies. Only one selected publication is provided for each technology.

Classification of rAST Detection of particular resistance mechanisms vs. universal phenotypic susceptibility testing Rapid detection of particular resistance mechanisms can be performed by DNA amplification methods targeting resistance genes or sequences, immunochromatographic assays or antibiotic degradation assays (Fig. 1). The last one can be accomplished as colorimetric [11] or matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS)-based [12] tests. An important limitation is that resistance detection is not the same as susceptibility testing, i.e. a negative result does not necessarily imply susceptibility. A number of alternative resistance mechanisms can still cause microbial resistance and, hence, treatment failure. For instance, despite lack of carbapenemase detection Gram-negative bacteria may be resistant due to reduced permeability or increased efflux [13e16]. In contrast, phenotypic susceptibility testing is universal, mechanism-independent and allows exact phenotypic categorization with direct therapeutic relevance. The inherent limitation of the growth-based phenotypic tests is that it takes time for the microorganisms to start (lag phase) and to perform (log phase) their growth and the expression of their response to the tested antibiotic agent [17]. The main pro and cons are given in Fig. 1. Since whole genome sequencing (WGS) allows the detection of a multitude of single resistance mechanisms, this approach offers the potential to predict antimicrobial susceptibility from a single assay. However, there are still many questions to be answered (Fig. 2). Genome-wide association studies, machine learning algorithms and detection of gene expression levels by RNA sequencing may

further improve the WGS approach for AST; however, they cannot replace universal phenotypic susceptibility testing [18]. According to the time to result, rAST has been defined as technologies yielding results in
8 hr [19], whereas ultra-rapid methods provide results in
4 hr [96]. This classification appears reasonable, because in addition to time and technological capabilities it reflects clinical purpose (see below) and laboratory workflow. According to clinical purpose, rAST assays can be divided in rapid single tests and accelerated full AST assays. Rapid single assays assess the activity of single antibiotics towards defined microorganisms in order to rapidly distinguish susceptible from resistant isolates under specific controlled conditions, testing, for example, meropenem vs. Enterobacterales members [20]. Single rapid tests can be used as add-on diagnostics when ordered ondemand in special, particularly in urgent situations. Applying those rapid single tests, full AST remains necessary for the isolate if assumed as clinically relevant. Owing to their add-on purpose, single rapid tests imply additional costs and work load. In contrast, accelerated full AST assays provide full susceptibility profile as with standard systems, i.e. complete and final susceptibility report of reasonable sets of multiple antimicrobials. Full AST methods, which allow acceleration compared to reference 18-hr testing, are available now with some automated systems [21]. However, the results for the whole panel of antibiotics are usually not available until evening or night hours if started during the standard working day. Given that most of the microbiology laboratories are not staffed during the evening and/or night-time hours for validation and forwarding the AST results to clinicians [22], even findings of the accelerated automated AST systems are usually only available at the next morning. At least for the near future, it appears

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

1349

Fig. 1. Detection of particular resistance mechanisms vs. universal phenotypic susceptibility testing. AST, antimicrobial susceptibility testing.

Fig. 2. Unsolved questions standing in the way of establishing whole genome sequencing (WGS) as approach for routine antimicrobial susceptibility testing (AST).

rather unrealistic that full AST results can be generated for multiple antibiotics very rapidly, i.e. within about 4 hr. Yet, it is very important to focus the technological advancement on AST results attainable within the same working shift, so that

they can be forwarded to clinicians within the same day. This would accelerate current AST diagnostics by one day with direct impact on patient management [6e8]. Technically, it seems quite realistic to generate full AST results within 8 hr [23], thus matching criteria for

1350

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

rapid testing as defined above. Nevertheless, it would be advantageous to further reduce the time of full rAST to about 6 hr, because set-up, validation and interpretation steps cost some additional time. As a time span of about 6 hr for the test procedures can be challenging for some antibiotics, decisions will have to be taken by assay developers about which antibiotics in which constellation are most needed for frequent and particularly urgent clinical circumstances with bacterial aetiology to still gain the benefit of result availability on the same day. MALDI-TOF MS, primarily introduced into routine diagnostics to accelerate identification, has also allowed more targeted, speciesspecific set-up for AST. Unlike earlier times, now species designation is usually known before susceptibility testing, and streamlined AST panels can be designed to comprise only relevant antibiotics. Thus, we are no longer so dependent on extremely broad AST panels that comprise numerous antibiotics to meet the requirements of testing widely varying microbial species. In this regard, it is reiterated that the overall time for the generation of a final report by the automatic AST systems depends on that antibiotic compound taking the longest time for testing.

action of these antimicrobials, namely killing activity against proliferating microorganisms [33e35]. Indeed, it has been demonstrated that the time to the onset of bacterial lysis is shorter in rapidly growing cells than in slowly growing cells [36]. The ability of a test system to differentiate between ‘growth’ or ‘no growth’ within 20 min of incubation does not necessarily mean that such a system is able to differentiate between resistance and susceptibility in such a short time. Early morphological alterations after the incubation start, such as, for example, filamentation in Gram-negative bacteria under b-lactams [32,37], create unspecific increase in growth curves. This leads to the superimposition of growth curves of those cultures with antibiotics with those of the growth control [37]. This obstacle hampers in particular optical methods [17,31,32,37] (Fig. S1), but may also affect other methods. Furthermore, systems based on other methods of early growth recognition still are hindered by the duration of the lag phase, during which the behaviour of microorganisms with or without antibiotics can hardly be distinguished with adequate reliability [20,38]. Delayed expression of resistance

Breakpoint tests vs. minimum inhibitory concentration (MIC) tests Breakpoint tests allow only categorization as susceptible and resistant by testing one breakpoint concentration, or as susceptible, intermediate and resistant by testing two breakpoint concentrations for organism/antibiotic combinations for which intermediate category is available in the interpretation guidelines. MIC tests include a broad range of antibiotic concentrations, thus providing not only interpretative categorization, but also exact MIC value. Availability of MIC allows fine adjustment of antimicrobial treatment based on pharmacokinetic/pharmacodynamic evaluation as well as sustainable epidemiological monitoring and data comparability even when breakpoints change over time. A strategic decision regarding development of an assay as breakpoint or MIC assay is taken after careful consideration of intended clinical use, technical issues and costs. Challenges of rAST and ways to overcome them Inoculum effect International guidelines recommend final inoculum size of 5  105 cfu/mL for the reference broth microdilution method [24,25] with an acceptable range of 2e8  105 cfu/mL. Decades ago it had already been demonstrated that deviations in inoculum size can lead to considerable discrepancy in AST result [26e28]. A significant inoculum effect was demonstrated when testing b-lactam antibiotics, particularly if b-lactamase production is the resistance mechanism [27,29]. A recent study showed that even minor differences in inoculum may have a dramatic effect on MIC determination [30]. Therefore, the inoculum effect can jeopardize the accuracy of technologies that do not utilize a standard inoculum size, e.g. some direct-from-specimen tests (see below). Lag phase and initial biomass increase in susceptible isolates While it is commonly assumed that very rapid growth detection by sensitive technologies is equal to the very rapid growth-based differentiation between susceptible and resistant isolates, in fact it does not hold true for all antibiotic/organism combinations. During the first hour(s) after the start of incubation, many susceptible organisms show initial biomass increase in the presence of b-lactam antibiotics before killing occurs [17,31,32] (see Fig. S1). This is most likely to be due to the peculiarity in the mechanism of

Usually, resistant isolates start to grow immediately or soon after a lag time in the presence of respective antibiotic, and their growth curve with antibiotic is similar to that of their growth control without antibiotic. However, some strains do not express their resistance from the very beginning, but behave as susceptible strains at the early time points of incubation [39,40]. During that time period, they cannot be differentiated from susceptible strains by phenotypic methods and would be misclassified as susceptible. After prolonged incubation, e.g. several hours, the resistance is expressed and the isolate starts to grow in spite of antibiotic. Possible explanations for the phenomenon of delayed phenotypic resistance are inducible resistance and the heteroresistance phenomenon [39e42]. Following issues regarding the impact of delayed resistance expression on rAST warrant further investigation. (i) What is the optimal incubation time for AST? Current reference methods measure endpoint (growth or no growth) after 18 ± 2 hr of incubation for the most organism/antibiotic combinations [24,25]. rAST assays must be compared to those reference methods and deviating finding is interpreted as an error of rapid assay. Thus, delayed resistance expression will result in a false-susceptible finding for a rapid method. However, the question arises of whether the current standard of 18 ± 2 hr is the best incubation time. If the absolute majority of isolates express their resistance sooner and this resistance is easily detectable, this standard incubation time may be considered as too conservative. On the other hand, resistance may still not be expressed until 18 ± 2 hr, further questioning the standard incubation time as an ultimate recommendation. Strains occur that carry a carbapenemase gene but being phenotypically susceptible to carbapenems even in a reference 18 ± 2 hr AST assay. While EUCAST recommends reporting such strains as susceptible to carbapenems [43], other experts warn that under circumstances this resistance can be expressed up to a level of clinical failure [44]. Experimental evidence suggests that high-level carbapenem resistance can be induced over time in carbapenemase-positive bacteria that initially exhibit the carbapenem susceptibility phenotype [39]. Another example is the varying expression of vanB-mediated vancomycin resistance in enterococci. It can be differently delayed depending on composition of medium and other test

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

1351

conditions in a rapid assay [40]. Nevertheless, it would be not helpful currently to change the standard incubation time for the reference assay, because the vast majority of data have been accumulated with the current conditions. The minimum incubation time required for rAST is organism/antibiotic specific and has to be assessed by developers in elaborate studies.

Otherwise, it can result in misinterpretation of deviating results as a failure of the test under development. Moreover, even the same nutrient media, e.g. cation-adjusted MuellereHinton broth or MuellereHinton agar from different manufacturers can provide considerably different AST results [47].

(ii) What proportion of clinical isolates shows delayed expression of resistance? Except for testing susceptibility of staphylococci towards oxacillin and vancomycin [42,45], this phenomenon seems to be uncommon in the case of usual ‘overnight’ AST, but its real incidence is unknown. This knowledge is important to understand how frequently this phenomenon would jeopardize rAST in general. (iii) What are the clinical consequences and risks? In particular, the risk should be considered of treating a patient with an isolate determined as susceptible by rAST due to delayed resistance, but tested resistant after standard incubation time in a standard assay. Is the risk of treatment failure (considerably) higher than with isolates being genotypically resistant, but phenotypically susceptible in a standard AST? Of note, this latter risk is accepted by current guidelines [43]. In the case of wrong result of a rapid test, there is still a possibility to adjust treatment on the next day when the results of the standard test e if performed in parallel e become available. It can be assumed that the initial growth inhibition observed in vitro may be reflected by the initial efficacy (at least to some degree) in vivo, but this issue necessitates detailed research. The transferability of the in vitro effect of brief action of antibiotics on bacteria to in vivo conditions has previously been pointed out as a research area of increasing clinical interest [37]. Generally, error risk due to delayed resistance expression should be better studied to be able to weigh up this potential risk of rAST against the important benefit of providing susceptibility information earlier. While accuracy of AST reporting remains the highest priority in diagnostic microbiology, it must be noted that too late arrival of a warning on multidrug-resistant organism will often have no consequences for the patient's survival.

The above considerations demonstrate that increased complexity and greater deviations of rAST technologies from the reference methods implicate higher risk of errors. Furthermore, the special challenges of rAST should be taken into account, which are limiting the rapidity of phenotypic AST, but not primarily the sensitivity of growth detection by different technologies e such as the initial biomass increase in susceptible isolates and the delayed expression of resistance. Focusing only on early recognition of growth in the development of novel rAST approaches will not be successful. Much more is required for decisive breakthrough and broad availability of rAST than merely sensitive detection of microbial growth. Key points for a successful rAST assay (Fig. 3) are discussed here in detail.

Standardization of testing conditions rAST inevitably implies deviations from the reference AST method, e.g. shortened incubation time or modified endpoint reading. However, it is advisable to keep the other test conditions as close as possible to the reference method guidelines, i.e. ISO [25], EUCAST [46] and CLSI [24]. Deviations from those recommended conditions, e.g. inoculum size, nutrient medium, breakpoint antibiotic concentrations, are critical. Since they hold the danger of reduction of the test accuracy, these deviations should be avoided unless really necessary for the test itself and investigated well. Choice of an appropriate reference method Broth microdilution, gradient diffusion and disk diffusion methods as standard AST methods can provide slightly or considerably different results, even when used in parallel for the same organism/antibiotic combination in strict compliance with the testing recommendations [47]. Therefore, during test development and validation it is advisable to choose the comparative reference method, which is most similar to the conditions of the rapid assay. For example, if the rapid assay under development is broth based, it is wise to use broth microdilution as the reference [24,25].

Criteria for a successful rAST assay

Capacity for simultaneous testing of multiple specimens and random access Many technologies in development and even some on the market share a common drawback of having limited capacities for simultaneous testing of multiple samples. At least a certain degree of high throughput is, however, an essential requirement if a machine is to be used in routine diagnostic laboratories. A ‘random access’ option, i.e. the feasibility of loading a further specimen while other samples are already in the testing process, is especially important for rAST approaches with their implied urgency of use. Affordability Expensive assays have low chances of being accepted. For acceptance and reimbursement of high-cost assays, good evidence quality will be needed, i.e. randomized controlled trials demonstrating positive effects on clinical outcomes. Such studies are rare and, again, increase the cost of assays. Expensive assays necessitate stringent indication criteria to choose the adequate patient population and/or situations for which the assay is performed. Such objective criteria are difficult to establish and even more difficult to apply in daily routine. Excessive costs prohibit assays from wide implementation and, as consequence, the patients will not benefit from this technology. Of note, rAST is also feasible with very basic and inexpensive methods, as exemplified below. Financial attractiveness for industry Decades ago, easy and cheap methods were suggested for rAST. Rapid growth-based methods are good examples, some of them accomplished as colorimetric assays [48e52]. While accurate results within a few hours were demonstrated, same-shift AST has still not become established for routine use. Recently, there has been revival of interest in such rapid colorimetric assays utilizing redox indicators [53,54] or pH indicators [55]. While it was demonstrated decades ago that results of the disk diffusion method can be read after very short incubation times [56e59], it was not until 2018 that EUCAST established standardized procedures and breakpoints for abbreviated incubation [60,61]. The published breakpoints are only authorized for rAST directly from positive blood culture (BC) bottles [61], while testing from colonies is not less urgently needed and feasible [62].

1352

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

Fig. 3. Challenges and requirements of rapid antimicrobial susceptibility testing (rAST).

As a substantial number of publications has demonstrated acceptable accuracy of those simple and cheap methods, it seems obvious that the main hurdles for their broad introduction into the routine are not of a technological nature. On the contrary, industry's willingness to develop these simple technologies to commercially available same-shift test products seems to be mainly influenced by limited financial interest. Investing in the industrial development of non-proprietary technologies entails a considerable financial risk. Indeed, a recent consensus statement has pointed out that protection of intellectual property is essential to make rAST technologies attractive for manufacturers [3]. Therefore, competition between commercial rAST assays should not primarily focus on minor differences in time to result, but rather on affordability, ease of use and integrability into the laboratory workflow. Direct AST from clinical specimens Direct testing from clinical specimens has an advantage of providing results 1 day earlier than the testing of isolates cultured overnight. Direct testing of positive BCs is particularly suitable due to its paramount clinical relevance, predominantly monomicrobial character and feasibility of inoculum standardization. Different centrifugation protocols yield a microbial pellet from positive BC broth, which is then inoculated in AST systems with or without prior inoculum standardization. The processing protocols vary between different studies and the reported accuracy results range from very good [31,63,64] to insufficient [65,66]. Lower accuracy was shown for Gram-positive cocci [66,67] and yeasts [65], particularly depending on the antimicrobial agent tested [65e67]. Besides revolutionary changes in microbial identification, MALDI-TOF MS has triggered rethinking in further areas of clinical microbiology, particularly in the acceleration of microbiological diagnostics to achieve more direct impact for patient care. After it has been realized that very short incubation on agar after subcultivation from positive BCs suffices for successful identification

by MALDI-TOF MS [68], the use of shortly cultured biomass on solid medium was also tried for AST. In a study, only 2.4 hr and 3.8 hr were enough to grow sufficient biomass of Gram-negative rods and Gram-positive cocci, respectively, from positive BCs to produce a standardized 0.5 McFarland turbidity suspension and inoculate an automated AST system [69]. The accuracy of this method was high with 99.2% categorical agreement compared with the inoculation of overnight cultures [69]. The high accuracy of the results is most probably due to the application of highly standardized inoculum from the grown culture, in contrast to the direct inoculation of microbial pellets [65,66]. AST from shortly incubated solid medium sub-cultures provides results 1 day earlier without any cost or time expenditure and can be optimally combined with the MALDI-TOF MS identification from the same ‘young’ biomass [15]. For most bacterial species, overnight incubation of sub-cultures is neither necessary for identification nor for AST. Taking into account that the results of automated AST systems will be only available on the next day in most settings, it appears reasonable to prefer inoculation from shortly incubated sub-cultures on solid medium over direct testing from microbial pellets in order to save reagents and hands-on time. Nevertheless, consideration can be given to direct processing without preceding short or long-term sub-cultivation, if pellet is used for rAST with the results being available on the same day. Mixed (polymicrobial) specimens Direct AST is feasible for specimen types that in most cases are expected to be monomicrobial and not contaminated by skin or mucosal microbiota. A prime example is positive BC, which is also particularly relevant for rapid testing taking into account the acute course of sepsis. Direct testing from urine is much more challenging to perform [70], as it is often polymicrobial or contaminated by colonizing microorganisms [9,71,72]. Therefore, practical utility of routine direct AST from urine samples may be questioned [9]. If performed, disk diffusion seems to be more suitable as this method can to some

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

degree distinguish polymicrobial growth [71e73]. Similar considerations apply for direct AST from respiratory samples [74e77]. The effect of inoculum and its impact on accuracy of AST has been discussed above. Since bacterial load in clinical specimens is not known beforehand, direct AST can be severely affected by the inoculum effect. Inoculum can be standardized prior to direct testing for some specimen types, e.g. positive BCs [31,63,64], whereas it is hardly feasible for other specimens types, such as respiratory or urine samples [71,72,74,75]. The gradient diffusion method is less prone to the inoculum effect [75,78], but its use for direct AST is limited by high cost when multiple antibiotics are to be tested [75]. MALDI-TOF MS-based susceptibility testing Introduction of MALDI-TOF MS has caused a revolution in clinical microbiology [79]. However, this radical change in routine diagnostics relates only to the identification of microorganisms, which has become more rapid and accurate. While being undoubtedly helpful, MALDI-TOF MS has caused some dissociation in the laboratory workflow. Previously, only one combined automated system had to be inoculated in a single step for both identification and AST, whereas now two instruments are separately inoculated. MALDI-TOF MS-based AST would therefore again enable the unification of those two basic functions of a microbiology laboratory in a single process. MALDI-TOF MS-based approaches comprise methods that allow either detection of particular resistance mechanisms or methods for universal AST. The microorganism-induced enzymatic modification of antibiotics can be detected by MALDI-TOF MS and has been described for b-lactamase production [12] as well as AAC (60 )-Ib-cr enzyme as a mechanism of fluoroquinolone resistance [80]. Another interesting approach is detection of a specific peak at 11 109 m/z, related to a pKpQIL plasmid carrying blaKPC in the MALDI-TOF mass spectra of KPC-producing Klebsiella pneumoniae [81]. Detecting only particular resistance mechanisms is the main inherent limitation of these MALDI-TOF MS-based approaches. A phenotypic AST by MALDI-TOF MS with short time to result has been proposed. This method, however, requires laborious processing including centrifugation and washing steps, as well as lysis of microbial cells in tube and subsequent transferral onto a MALDI target [82]. Recently, a novel direct-on-target microdroplet growth assay (DOT-MGA) has been suggested as a rapid universal MALDI-TOF MS-based AST method [20]. This phenotypic method is easy to perform as it utilizes application and incubation of microbial suspension as microdroplets with and without antibiotic directly on MALDI targets, followed by broth removal. After MALDI-TOF MS measurement, the results are evaluated by a simple algorithm. The method is independent of resistance mechanism and close to the CLSI and ISO broth microdilution methods, which contributes to the high agreement with this reference method [20]. The method can be expanded to further applications, such as susceptibility determination directly from clinical specimens [63] and simultaneous testing of multiple antibiotics [23]. It also has the capability for automation to allow high-throughput testing. For detailed reading on MALDI-TOF MS-based AST methods, we refer to recent review articles [83,84]. Optimization of workflow and infrastructural issues Rapid availability of AST results is limited not only by technological reasons, but also by organizational aspects, the latter being at least just as important [85]. A recent survey of 209 laboratories in 25 European countries revealed that only 42.2% of laboratories

1353

were able to start blood cultivation in automated BC systems around the clock and only 13% of laboratories had a 24-hour service to start processing of positive BC bottles [22]. Less than 5% of laboratories could validate and transmit the identification and AST findings for BC pathogens to the clinicians 24 hr/day; 21.6% of laboratories generally do not forward identification and AST results to the clinicians on Sundays [22]. More than 30 years ago, at the time of introduction of semiautomated AST systems into microbiological routine diagnostics, critics warned of failed expectation of greater speed [32]. Inflexible time schedules of hospitals and staff function were identified as the major constraint to achieving more rapid availability and dissemination of laboratory results [32]. Ignoring these fundamental organizational aspects [86] will counteract any race for the development of highly sophisticated (and even more expensive) AST platforms. In conclusion, joint efforts have to be made by industry, authority and academia to enable swift and broad introduction of rAST into clinical diagnostics. Optimization of laboratory organization and workflow are essential in addition to technology improvement. Only affordable and easy-to-use rAST methods will have the chance to become widely accepted.

Transparency declaration E.A.I. and K.B. are inventors of patent applications in the field of diagnostic microbiology, owned by the University of Münster, including a patent application on rAST licensed to Bruker Daltonik. E.A.I. received speaker honorarium from Bruker Daltonik. K.B. received lecture fees from Becton Dickinson, Bruker Daltonik, Hain Lifescience, Roche Diagnostics and ThermoFisher. This work was partly funded by a grant from the German Federal Ministry of Education and Research (BMBF) to E.A.I and K.B. (16GW0150).

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.cmi.2019.04.025.

References €gberg LD, Plachouras D, Quattrocchi A, Hoxha A, Simonsen GS, [1] Cassini A, Ho et al. Attributable deaths and disability-adjusted life-years caused by infections with antibiotic-resistant bacteria in the EU and the European Economic Area in 2015: a population-level modelling analysis. Lancet Infect Dis 2019;19:56e66. [2] Roca I, Akova M, Baquero F, Carlet J, Cavaleri M, Coenen S, et al. The global threat of antimicrobial resistance: science for intervention. New Microbe New Infect 2015;6:22e9. [3] van Belkum A, Bachmann TT, Ludke G, Lisby JG, Kahlmeter G, Mohess A, et al. Developmental roadmap for antimicrobial susceptibility testing systems. Nat Rev Microbiol 2019;17:51e62. [4] Kumar A, Ellis P, Arabi Y, Roberts D, Light B, Parrillo JE, et al. Initiation of inappropriate antimicrobial therapy results in a fivefold reduction of survival in human septic shock. Chest 2009;136:1237e48. [5] Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock: 2016. Intensive Care Med 2017;43:304e77. [6] Barenfanger J, Drake C, Kacich G. Clinical and financial benefits of rapid bacterial identification and antimicrobial susceptibility testing. J Clin Microbiol 1999;37:1415e8. [7] Doern GV, Vautour R, Gaudet M, Levy B. Clinical impact of rapid in vitro susceptibility testing and bacterial identification. J Clin Microbiol 1994;32: 1757e62. [8] Kerremans JJ, Verboom P, Stijnen T, Hakkaart-van RL, Goessens W, Verbrugh HA, et al. Rapid identification and antimicrobial susceptibility testing reduce antibiotic use and accelerate pathogen-directed antibiotic use. J Antimicrob Chemother 2008;61:428e35. [9] Stratton CW. Advanced phenotypic antimicrobial susceptibility testing methods. In: Tang YW, Stratton CW, editors. Advanced techniques in diagnostic microbiology. Cham: Springer; 2018. p. 69e98.

1354

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355

[10] Leonard H, Colodner R, Halachmi S, Segal E. Recent advances in the race to design a rapid diagnostic test for antimicrobial resistance. ACS Sens 2018;3: 2202e17. [11] Nordmann P, Poirel L, Dortet L. Rapid detection of carbapenemase-producing Enterobacteriaceae. Emerg Infect Dis 2012;18:1503e7. [12] Burckhardt I, Zimmermann S. Using matrix-assisted laser desorption ionization-time of flight mass spectrometry to detect carbapenem resistance within 1 to 2.5 hours. J Clin Microbiol 2011;49:3321e4. [13] Chalhoub H, Saenz Y, Rodriguez-Villalobos H, Denis O, Kahl BC, Tulkens PM, et al. High-level resistance to meropenem in clinical isolates of Pseudomonas aeruginosa in the absence of carbapenemases: role of active efflux and porin alterations. Int J Antimicrob Agents 2016;48:740e3. [14] Culbreath K, Petti CA. Balancing enthusiasm for innovative technologies with optimizing value: an approach to adopt new laboratory tests for infectious diseases using bloodstream infections as exemplar. Open Forum Infect Dis 2015;2:ofv075. [15] Idelevich EA, Becker K. Identification and susceptibility testing from shortly incubated cultures accelerate blood culture diagnostics at no cost. Clin Infect Dis 2016;62:268e9. [16] Dien Bard J, Lee F. Why can't we just use PCR? The role of genotypic versus phenotypic testing for antimicrobial resistance testing. Clin Microbiol News 2018;40:87e95. €rlich D, Knaack D, Grünastel B, Peters G, et al. Rapid [17] Idelevich EA, Hoy M, Go phenotypic detection of microbial resistance in Gram-positive bacteria by a real-time laser scattering method. Front Microbiol 2017;8:1064. [18] Ellington MJ, Ekelund O, Aarestrup FM, Canton R, Doumith M, Giske C, et al. The role of whole genome sequencing in antimicrobial susceptibility testing of bacteria: report from the EUCAST Subcommittee. Clin Microbiol Infect 2017;23:2e22. [19] Buehler SS, Madison B, Snyder SR, Derzon JH, Cornish NE, Saubolle MA, et al. Effectiveness of practices to increase timeliness of providing targeted therapy for inpatients with bloodstream infections: a laboratory medicine best practices systematic review and meta-analysis. Clin Microbiol Rev 2016;29:59e103. [20] Idelevich EA, Sparbier K, Kostrzewa M, Becker K. Rapid detection of antibiotic resistance by MALDI-TOF mass spectrometry using a novel direct-on-target microdroplet growth assay. Clin Microbiol Infect 2018;24:738e43. [21] Eigner U, Schmid A, Wild U, Bertsch D, Fahr AM. Analysis of the comparative workflow and performance characteristics of the VITEK 2 and Phoenix systems. J Clin Microbiol 2005;43:3829e34. [22] Idelevich EA, Seifert H, Sundqvist M, Scudeller L, Amit S, Balode A, et al. Microbiological diagnostics of bloodstream infections in Europe-an ESGBIES survey. Clin Microbiol Infect 2019;25:1399e407. https://doi.org/10.1016/ j.cmi.2019.03.024. [23] Idelevich EA, Busch JA, Sparbier K, Drews O, Kostrzewa M, Becker K. Rapid simultaneous testing of multiple antibiotics by the MALDI-TOF MS-based direct-on-target microdroplet growth assay (DOT-MGA). O0745. 28th ECCMID. Madrid, Spain: ESCMID Online Lecture Library; 2018. [24] CLSI. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; approved standard. 11th ed. Wayne, PA: Clinical Laboratory Standards Institute; 2018. M07-A11. [25] ISO. 20776e1. Clinical laboratory testing and in vitro diagnostic test systems susceptibility testing of infectious agents and evaluation of performance of antimicrobial susceptibility test devices - Part 1: reference method for testing the in vitro activity of antimicrobial agents against rapidly growing aerobic bacteria involved in infectious diseases. Geneva, Switzerland: International Organization for Standardization; 2006. [26] Brook I. Inoculum effect. Rev Infect Dis 1989;11:361e8. [27] Sabath LD, Garner C, Wilcox C, Finland M. Effect of inoculum and of betalactamase on the anti-staphylococcal activity of thirteen penicillins and cephalosporins. Antimicrob Agents Chemother 1975;8:344e9. [28] Luria SE. A test for penicillin sensitivity and resistance in Staphylococcus. Proc Soc Exp Biol Med 1946;61:46e51. [29] Adler A, Ben-Dalak M, Chmelnitsky I, Carmeli Y. Effect of resistance mechanisms on the inoculum effect of carbapenem in Klebsiella pneumoniae isolates with borderline carbapenem resistance. Antimicrob Agents Chemother 2015;59:5014e7. [30] Smith KP, Kirby JE. The inoculum effect in the era of multidrug resistance: minor differences in inoculum have dramatic effect on MIC determination. Antimicrob Agents Chemother 2018;62. e00433e18. € rlich D, Peters G, Borowski M, et al. Direct [31] Idelevich EA, Hoy M, Knaack D, Go determination of carbapenem-resistant Enterobacteriaceae and Pseudomonas aeruginosa from positive blood cultures using laser scattering technology. Int J Antimicrob Agents 2018;51:221e6. [32] Cherubin CE, Eng R, Appleman M. A critique of semiautomated susceptibility systems. Rev Infect Dis 1987;9:655e9. [33] Tuomanen E, Cozens R, Tosch W, Zak O, Tomasz A. The rate of killing of Escherichia coli by ß-lactam antibiotics is strictly proportional to the rate of bacterial growth. J Gen Microbiol 1986;132:1297e304. [34] Hobby GL, Meyer K, Chaffee E. Observations on the mechanism of action of penicillin. Proc Soc Exp Biol Med 1942;50:281e5. [35] Hobby GL, Dawson MH. Effect of rate of growth of bacteria on action of penicillin. Proc Soc Exp Biol Med 1944;56:181e4. [36] Boman HG, Eriksson KG. Penicillin-induced lysis in Escherichia coli. J Gen Microbiol 1963;31:339e52.

[37] Yourassowsky E, van der Linden MP, Lismont MJ, Crokaert F, Glupczynski Y. Correlation between growth curve and killing curve of Escherichia coli after a brief exposure to suprainhibitory concentrations of ampicillin and piperacillin. Antimicrob Agents Chemother 1985;28:756e60. [38] Rolain JM, Mallet MN, Fournier PE, Raoult D. Real-time PCR for universal antibiotic susceptibility testing. J Antimicrob Chemother 2004;54:538e41. [39] Adams-Sapper S, Nolen S, Donzelli GF, Lal M, Chen K, Justo da Silva LH, et al. Rapid induction of high-level carbapenem resistance in heteroresistant KPCproducing Klebsiella pneumoniae. Antimicrob Agents Chemother 2015;59: 3281e9. [40] Jett B, Free L, Sahm DF. Factors influencing the Vitek gram-positive susceptibility system's detection of vanB-encoded vancomycin resistance among enterococci. J Clin Microbiol 1996;34:701e6. [41] Kayser FH, Benner EJ, Hoeprich PD. Acquired and native resistance of Staphylococcus aureus to cephalexin and other ß-lactam antibiotics. Appl Microbiol 1970;20:1e5. €ssle A, Kayser FH, Berger-Ba €chi B. Mechanisms of hetero[42] Ryffel C, Stra resistance in methicillin-resistant Staphylococcus aureus. Antimicrob Agents Chemother 1994;38:724e8. [43] EUCAST. Breakpoint tables for interpretation of MICs and zone diameters. The European Committee on Antimicrobial Susceptibility Testing; 2018. Version 8.1, www.eucast.org. [44] Livermore DM, Andrews JM, Hawkey PM, Ho PL, Keness Y, Doi Y, et al. Are susceptibility tests enough, or should laboratories still seek ESBLs and carbapenemases directly? J Antimicrob Chemother 2012;67:1569e77. [45] Sieradzki K, Villari P, Tomasz A. Decreased susceptibilities to teicoplanin and vancomycin among coagulase-negative methicillin-resistant clinical isolates of staphylococci. Antimicrob Agents Chemother 1998;42:100e7. [46] EUCAST. EUCAST disk diffusion method for antimicrobial susceptibility testing. The European Committee on Antimicrobial Susceptibility Testing; 2017. Version 6.0, www.eucast.org. [47] Idelevich EA, Büsing M, Mischnik A, Kaase M, Bekeredjian-Ding I, Becker K. False non-susceptible results of tigecycline susceptibility testing against Enterobacteriaceae by an automated system: a multicentre study. J Med Microbiol 2016;65:877e81. [48] Pital A, Disque DT, Leise JM. A new rapid plate method for determining antibiotic sensitivity. Antibiot Chemother (Northfield) 1956;6:351e9. [49] Rajam PC, Adcock JD. A rapid test for screening the sensitivity of staphylococci to antibiotics. Am J Clin Pathol 1953;23:1168e72. [50] Rogers MA, Ryan WL, Severens JM. A new method for the rapid determination of bacterial sensitivity. Antibiot Chemother (Northfield) 1955;5:382e5. [51] Schneierson SS. A simple rapid disk-tube method for determination of bacterial sensitivity to antibiotics. Antibiot Chemother (Northfield) 1954;4: 125e32. [52] Sorensen RH. Rapid antibiotic sensitivity test using a redox indicator. Med Techn Bull 1959;10:144e50. [53] Lescat M, Poirel L, Tinguely C, Nordmann P. A resazurin reduction-based assay for rapid detection of polymyxin resistance in Acinetobacter baumannii and Pseudomonas aeruginosa. J Clin Microbiol 2019;57. e01563-18. [54] March-Rossello GA, Gutierrez-Rodriguez MP, Simarro-Grande M, OrdunaDomingo A, Bratos-Perez MA. A two-hour procedure for determining the susceptibility of enterococci and staphylococci to antibiotics by a colourimetric method. Rev Esp Quimioter 2015;28:247e55. [55] Nordmann P, Jayol A, Poirel L. Rapid detection of polymyxin resistance in enterobacteriaceae. Emerg Infect Dis 2016;22:1038e43. [56] Barry AL, Joyce LJ, Adams AP, Benner EJ. Rapid determination of antimicrobial susceptibility for urgent clinical situations. Am J Clin Pathol 1973;59: 693e9. [57] Coyle MB, McGonagle LA, Plorde JJ, Clausen CR, Schoenknecht FD. Rapid antimicrobial susceptibility testing of isolates from blood cultures by direct inoculation and early reading of disk diffusion tests. J Clin Microbiol 1984;20: 473e7. [58] Kluge RM. Accuracy of Kirby-Bauer susceptibility tests read at 4, 8, and 12 hours of incubation: comparison with readings at 18 to 20 hours. Antimicrob Agents Chemother 1975;8:139e45. [59] Liberman DF, Robertson RG. Evaluation of a rapid Bauer-Kirby antibiotic susceptibility determination. Antimicrob Agents Chemother 1975;7:250e5. [60] EUCAST. EUCAST rapid AST directly from positive blood culture bottles. 2018. Version 1.0, www.eucast.org. [61] EUCAST. Zone diameter breakpoints for rapid antimicrobial susceptibility testing (RAST) directly from blood culture bottles. 2018. Version 1.0, www. eucast.org. € chliger N, Kolesnik-Goldmann N, Bo €ttger EC. Fully [62] Hombach M, Jetter M, Blo automated disc diffusion for rapid antibiotic susceptibility test results: a proof-of-principle study. J Antimicrob Chemother 2017;72:1659e68. [63] Idelevich EA, Storck LM, Sparbier K, Drews O, Kostrzewa M, Becker K. Rapid direct susceptibility testing from positive blood cultures by the matrixassisted laser desorption ionization-time of flight mass spectrometry-based direct-on-target microdroplet growth assay. J Clin Microbiol 2018;56: e00913e8. [64] Prod'hom G, Durussel C, Greub G. A simple blood-culture bacterial pellet preparation for faster accurate direct bacterial identification and antibiotic susceptibility testing with the VITEK 2 system. J Med Microbiol 2013;62: 773e7.

E.A. Idelevich, K. Becker / Clinical Microbiology and Infection 25 (2019) 1347e1355 [65] Idelevich EA, Grunewald CM, Wüllenweber J, Becker K. Rapid identification and susceptibility testing of Candida spp. from positive blood cultures by combination of direct MALDI-TOF mass spectrometry and direct inoculation of Vitek 2. PLoS One 2014;9:e114834. [66] de Cueto M, Ceballos E, Martinez-Martinez L, Perea EJ, Pascual A. Use of positive blood cultures for direct identification and susceptibility testing with the Vitek 2 system. J Clin Microbiol 2004;42:3734e8. [67] Kerremans JJ, Goessens WH, Verbrugh HA, Vos MC. Accuracy of identification and susceptibility results by direct inoculation of Vitek 2 cards from positive BACTEC cultures. Eur J Clin Microbiol Infect Dis 2004;23:892e8. [68] Idelevich EA, Schüle I, Grünastel B, Wüllenweber J, Peters G, Becker K. Rapid identification of microorganisms from positive blood cultures by MALDI-TOF mass spectrometry subsequent to very short-term incubation on solid medium. Clin Microbiol Infect 2014;20:1001e6. [69] Idelevich EA, Schüle I, Grünastel B, Wüllenweber J, Peters G, Becker K. Acceleration of antimicrobial susceptibility testing of positive blood cultures by inoculation of Vitek 2 cards with briefly incubated solid medium cultures. J Clin Microbiol 2014;52:4058e62. [70] Veron L, Mailler S, Girard V, Muller BH, L'Hostis G, Ducruix C, et al. Rapid urine preparation prior to identification of uropathogens by MALDI-TOF MS. Eur J Clin Microbiol Infect Dis 2015;34:1787e95. [71] Breteler KB, Rentenaar RJ, Verkaart G, Sturm PD. Performance and clinical significance of direct antimicrobial susceptibility testing on urine from hospitalized patients. Scand J Infect Dis 2011;43:771e6. [72] Sundqvist M, Olafsson J, Matuschek E. EUCAST breakpoints can be used to interpret direct susceptibility testing of Enterobacteriaceae from urine samples. APMIS 2015;123:152e5. [73] Zboromyrska Y, Rubio E, Alejo I, Vergara A, Mons A, Campo I, et al. Development of a new protocol for rapid bacterial identification and susceptibility testing directly from urine samples. Clin Microbiol Infect 2016;22:561e6. [74] Kontopidou F, Galani I, Panagea T, Antoniadou A, Souli M, Paramythiotou E, et al. Comparison of direct antimicrobial susceptibility testing methods for rapid analysis of bronchial secretion samples in ventilator-associated pneumonia. Int J Antimicrob Agents 2011;38:130e4. [75] Le Dorze M, Gault N, Foucrier A, Ruppe E, Mourvillier B, Woerther PL, et al. Performance and impact of a rapid method combining mass spectrometry and direct antimicrobial susceptibility testing on treatment adequacy of patients with ventilator-associated pneumonia. Clin Microbiol Infect 2015;21. 468.e16. [76] Perez LR, Freitas AL, Barth AL, Dias CA. Direct disk diffusion susceptibility testing from respiratory tract specimens: focus on Pseudomonas aeruginosa. Int J Infect Dis 2014;26:47e8. [77] Shah PM. Direct susceptibility testing of respiratory samples. Clin Infect Dis 2007;45:401e3. [78] Cercenado E, Cercenado S, Marin M, Rico MV, Vicente T, Bouza E. Evaluation of direct E-test on lower respiratory tract samples: a rapid and accurate procedure for antimicrobial susceptibility testing. Diagn Microbiol Infect Dis 2007;58:211e6. [79] Bizzini A, Greub G. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry, a revolution in clinical microbial identification. Clin Microbiol Infect 2010;16:1614e9.

1355

[80] Pardo CA, Tan RN, Hennequin C, Beyrouthy R, Bonnet R, Robin F. Rapid detection of AAC(6')-Ib-cr production using a MALDI-TOF MS strategy. Eur J Clin Microbiol Infect Dis 2016;35:2047e51. [81] Lau AF, Wang H, Weingarten RA, Drake SK, Suffredini AF, Garfield MK, et al. A rapid matrix-assisted laser desorption ionization-time of flight mass spectrometry-based method for single-plasmid tracking in an outbreak of carbapenem-resistant Enterobacteriaceae. J Clin Microbiol 2014;52: 2804e12. [82] Lange C, Schubert S, Jung J, Kostrzewa M, Sparbier K. Quantitative matrixassisted laser desorption ionization-time of flight mass spectrometry for rapid resistance detection. J Clin Microbiol 2014;52:4155e62. [83] Burckhardt I, Zimmermann S. Susceptibility testing of bacteria using MaldiTof mass spectrometry. Front Microbiol 2018;9:1744. [84] Oviano M, Bou G. Matrix-assisted laser desorption ionization-time of flight mass spectrometry for the rapid detection of antimicrobial resistance mechanisms and beyond. Clin Microbiol Rev 2019;32:e00037e18. [85] Idelevich EA, Reischl U, Becker K. New microbiological techniques in the diagnosis of bloodstream infections. Dtsch Arztebl Int 2018;115:822e32. [86] Blondeau JM, Idelevich EA. The 24-h clinical microbiology service is essential for patient management. Future Microbiol 2018;13:1625e8. [87] Choi J, Jung YG, Kim J, Kim S, Jung Y, Na H, et al. Rapid antibiotic susceptibility testing by tracking single cell growth in a microfluidic agarose channel system. Lab Chip 2013;13:280e7. [88] Fredborg M, Andersen KR, Jorgensen E, Droce A, Olesen T, Jensen BB, et al. Real-time optical antimicrobial susceptibility testing. J Clin Microbiol 2013;51:2047e53. [89] Godin M, Delgado FF, Son S, Grover WH, Bryan AK, Tzur A, et al. Using buoyant mass to measure the growth of single cells. Nat Methods 2010;7:387e90. [90] von Ah U, Wirz D, Daniels AU. Rapid differentiation of methicillin-susceptible Staphylococcus aureus from methicillin-resistant S. aureus and MIC determinations by isothermal microcalorimetry. J Clin Microbiol 2008;46: 2083e7. €jer H, Ansehn S, Bro €te L. Effects of ampicillin on intra[91] Thore A, Nilsson L, Ho cellular levels of adenosine triphosphate in bacterial cultures related to antibiotic susceptibility. Acta Pathol Microbiol Scand B 1977;85:161e6. [92] Riska PF, Su Y, Bardarov S, Freundlich L, Sarkis G, Hatfull G, et al. Rapid filmbased determination of antibiotic susceptibilities of Mycobacterium tuberculosis strains by using a luciferase reporter phage and the Bronx Box. J Clin Microbiol 1999;37:1144e9. [93] Descours G, Desmurs L, Hoang TLT, Ibranosyan M, Baume M, Ranc AG, et al. Evaluation of the Accelerate Pheno™ system for rapid identification and antimicrobial susceptibility testing of Gram-negative bacteria in bloodstream infections. Eur J Clin Microbiol Infect Dis 2018;37:1573e83. [94] Leonard H, Halachmi S, Ben-Dov N, Nativ O, Segal E. Unraveling antimicrobial susceptibility of bacterial networks on micropillar architectures using intrinsic phase-shift spectroscopy. ACS Nano 2017;11:6167e77. [95] Pore RS. Antibiotic susceptibility testing of Candida albicans by flow cytometry. Curr Microbiol 1990;20:323e8. [96] Kahlmeter G. Rapid phenotypic susceptibility testing. S536. 26th ECCMID 2016. Amsterdam, The Netherlands: ESCMID Online Lecture Library; 2016.