Table 1.

Current rapid methods/methods under development for Gram-negative resistance typing

	Targeted resistance mechanism	Targeted population include	Commercial systems include	Typical turnaround time	Performance	Limitations include
MALDI-TOF31,32,43–46	All antibiotics. Genotypic equivalent detects either specific protein fragmentation peak patterns (compared with a data library) or biochemical equivalent detects hydrolysis of antibiotics, uptake of stable isotopes in presence of antibiotic or bacterial growth in presence/absence of antibiotics with internal standard compound	All patients with serious/life-threatening infections/sepsis requiring urgent treatment including BSIs, meningitis	MALDI-TOF resistance typing (Bruker Biotyper® and VITEK® MS)	1–4 ha (with some methods this is dependent on incubation time with antibiotic)	Sensitivity: 80%–100%	High upfront costs due to sophisticated hardware and complex databases/data processing involved.22,32
					Specificity: 90%–100% depending on bacterial species and resistance type
					MALDI-TOF can be more effective as part of an AMS programme	Further optimization is needed for detection of resistance amongst various pathogens32 and for differentiation of certain bacterial species.45 Continuous upgrades to the databases and optimal sample enrichment will increase its accuracy45
Colorimetric tests22,31,40,47,48	All antibiotics—susceptibility/resistance detected by multiple different means. Colorimetric approaches include phenotyping, detection of bacterial growth (volatile compounds), degradation of the antibiotic and detection of specific resistance genes	All patients with serious or less serious infections	VITEK® 2XL, BD Phoenix™, Beckman Microscan, Sensititre™ Aris™ 2X	Methods requiring bacterial growth (minimum 12 h, can be as long as 30–40 h in total); nucleic acid amplification methods are more rapid	Sensitivity: 95%	Some colorimetric tests require culture and can be slow.48 Also, some colorimetric tests have a narrow antibiotic range and limited panel capacity, can be expensive, and require at least ∼105 cells22
					Specificity: 98%–100% (nucleic acid methods can be more variable)
FISH34,49–52	Selected antibiotic resistance—mostly involving ribosomal changes, e.g. clarithromycin and linezolid. FISH has been used for the detection of ESBLs. Frequently used for Helicobacter and Campylobacter species	Patients with various infections such as gastrointestinal, BSIs and respiratory tract infections		60–90 min—some methods are faster	Variable—sensitivity usually reported to be 80%–100% depending on bacterial species (≥90% sensitivity and >98% specificity reported for K. pneumoniae and P. aeruginosa)	Requires great skill and experience, can have low sensitivity compared with PCR, the organism(s) causing the infection have to be anticipated before the probes are chosen (requires a structured diagnostic algorithm), the density of pathogens should be ≥100 000 cfu/mL51 and FISH probe panels need to be tailored to individual needs52
FISH + time-lapse and automated photography49,53,54	Time-lapse photography over seconds after photobleaching transiently reveals the presence of certain bacterial species after hybridization with nucleic acid mimics. Used in detecting bacteria and their antibiotic resistance genes in patient samples, e.g. gut mucus	Patients with specific infections such as Helicobacter pylori		30–90 min (mostly hybridization time)	Not specified in reports identified	The ability of nucleic acid probes to hybridize efficiently can be hindered by the presence of mucus in the samples54
Molecular detection systems (nucleic acid amplification-based)30,55,56	Multiple different antibiotic resistance mechanisms including DNA gyrase, ribosomal and PBP mutations, BLs, ESBLs, carbapenemases, and membrane pump and permeability/porin-related resistance	Rapid detection of colonized patients and healthcare workers	GeneXpert, (Cepheid—with different kits e.g. Xpert® Carba-R), Check-MDR (Check-Points), BD MAX™ (Becton Dickinson)	1–3 h	Sensitivity: 73%–100%	PCR requires a high copy number of the target gene.30 Certain tests (e.g. POC mPCR) can have a high rate of complete or partial test failures, leading to non-concordant results in up to 45% of cases57
					Specificity: 90.5%–94.5% (for carbapenemases)	False-positive results can arise from residual DNA from dead bacteria, or by detecting bacteria harbouring (but not expressing) certain genes31
DNA microarray49	Based on hybridization to detect resistant bacteria in samples such as blood and respiratory specimens. Uses an array of gene sequences for multiple antimicrobial resistance markers, including those for BLs and ESBLs	Patients with various infections	Verigene® Nanosphere, SeptiCyte®, VAPChip	2.5–8 h	Sensitivity: 72.9%	Microarrays are still considered too complex and protracted for routine use in the clinic, and the range of organisms that can be detected is limited
					Specificity: 99.1% (for VAPChip assay only)	They are also subject to risk of contamination and are expensive to run49
NG-Test® CARBA 5 immunochromatographic test58,59	The qualitative test CARBA 5 will detect the five most common carbapenemase families (KPC, OXA-48-like, VIM, IMP and NDM) directly from Enterobacterales and P. aeruginosa bacterial colonies	For use as an infection control aid in the detection of carbapenemase-producers in healthcare settings		≤15 mina	Comparison with composite reference method:	Requires overnight culture, it has only been validated with certain types of agar and with Enterobacterales and P. aeruginosa. Organism identification is required prior to testing, and further validation across different sample types (e.g. blood and urine) is needed58,59
					PPA: 98.9%–100%
					NPA: 95.2%–100%
Clinical metagenomics42	NGS of nucleic acids isolated from clinical samples is performed to detect all microbes simultaneously	Intended for rapid and unbiased pathogen identification in clinical specimens	Nanopore sequencing platform (Oxford Nanopore)	6–8 h	Sensitivity: 96.6%	False positives are possible, which may need additional radiographic/clinical investigation
					Specificity: 88.0%
					PPV: 92.3%	Incurs much greater costs than traditional methods42
					NPV: 94.5% (sputum and BAL samples)

BL, β-lactamase; FISH, fluorescence in-situ hybridization; NPA, negative percentage agreement; POC, point-of-care; PPA, positive percentage agreement; PPV, positive predictive value.

^aDoes not include the time required if an initial bacterial culture is needed before sample processing.