Antimicrobial Susceptibility Testing and Resistance Mechanisms

Antimicrobial susceptibility testing (AST) exposes an isolated pathogen to an antimicrobial agent and evaluates growth inhibition or killing. Growth inhibition is the routine endpoint; killing is reserved for narrow bactericidal-testing indications. In the United States, the Clinical and Laboratory Standards Institute (CLSI) publishes the consensus standards governing media, inoculum, incubation, quality control, and breakpoint interpretation used by US clinical laboratories: CLSI M02 for disk diffusion, M07 for dilution methods, M11 for anaerobes, and M100 for the interpretive breakpoint and quality-control tables, currently in its 36th edition, M100-Ed36, published January 2026. The July 9, 2026 correction notice restores the sulfamethoxazole concentrations in the trimethoprim-sulfamethoxazole MIC categories in Table 2H-1 for Streptococcus spp. β-hemolytic group. Those corrected MIC categories apply to skin and skin-structure isolates. CLSI revises M100 annually, and breakpoints for a given drug-organism combination can shift between editions; a laboratory must interpret and report against the specific M100 edition it has adopted and validated, and must update its methods and expert-system rules each time it adopts a new edition. Exact numeric zone and MIC tables other than the publicly described S. aureus vancomycin categories and the Table 2H-1 correction remain paywalled and are not quoted here.¹

Mycobacterial and Nocardia susceptibility procedures are covered in mycobacteriology.

Principles, media, inoculum, and incubation

Mueller-Hinton agar or broth is the default medium because it has good lot-to-lot reproducibility, is low in sulfonamide/trimethoprim/tetracycline inhibitors, and supports most nonfastidious bacteria. Cation content directly affects results. Excess calcium and magnesium make Pseudomonas aeruginosa falsely appear more resistant to aminoglycosides. Fastidious organisms require supplementation or a different medium. Streptococci need 5% sheep blood for disk diffusion or 2.5-5% lysed horse blood for broth dilution; Haemophilus influenzae requires Haemophilus test medium; N. gonorrhoeae requires GC agar base with growth supplement; and the B. fragilis group is tested in supplemented Brucella broth. Agar for disk diffusion must be 4 mm deep. Agar that is too deep falsely increases apparent resistance because less antibiotic reaches a given radius, whereas agar that is too shallow falsely increases apparent susceptibility.

A standardized inoculum equivalent to a 0.5 McFarland turbidity standard (approximately 1-2 × 10⁸ CFU/mL) is required, prepared either by direct colony suspension from an 18-24 hour culture or, for fastidious/hard-to-suspend organisms, by growth-based standardization in broth.

General incubation conditions are 35 °C ± 2 °C in ambient air for 16-24 hours, but numerous organism-specific exceptions exist: Streptococcus spp. and Haemophilus spp. require 5% CO₂ for disk diffusion, and Haemophilus zones are read at 16-18 hours while Streptococcus zones require 20-24 hours before reading.

Interpretation: MIC, breakpoints, and categories

The minimum inhibitory concentration (MIC) is the lowest antimicrobial concentration that prevents visible growth and is the basic quantitative result of most phenotypic AST. An MIC has little clinical meaning without a breakpoint, a clinically validated concentration threshold that sorts isolates into categories predicting therapeutic success. CLSI defines the following interpretive categories:

Interpretive categories predict the likelihood of therapeutic success at the stated breakpoint. Site of infection can change the category meaning: a β-lactam that concentrates in urine may still be useful when the isolate tests intermediate.

Dilution and gradient-diffusion methods

Application: broth microdilution and MIC/breakpoint interpretation. An isolate is tested with a doubling series of Drug Y at 0.25, 0.5, 1, 2, 4, 8, 16, and 32 µg/mL. Visible growth occurs through 2 µg/mL, while the 4-µg/mL well and all higher concentrations remain clear. The MIC is therefore 4 µg/mL. If the applicable breakpoints are susceptible at 2 µg/mL or lower, intermediate at 4 µg/mL, and resistant at 8 µg/mL or higher, the isolate is reported intermediate.

Disk diffusion

The disk diffusion (Kirby-Bauer) method is limited to rapidly growing aerobic and facultative bacteria. A paper disk containing a fixed amount of drug is placed on an agar surface streaked with a standardized inoculum. The drug diffuses outward, and the diameter of the resulting zone of no growth correlates inversely with the MIC. A single 150-mm plate accommodates up to about 12 antibiotic disks. The method is inexpensive, simple to modify for local antibiogram needs, and provides results that are easily communicated to clinicians. It is manual and qualitative, providing only a category rather than an MIC, so its throughput becomes limiting in high-volume laboratories.³

Direct β-lactamase testing is a related but distinct phenotypic method: a chromogenic-cephalosporin (nitrocefin) disk changes color on hydrolysis by a penicillinase, giving a rapid same-day answer for β-lactamase production (useful in Haemophilus influenzae and enterococci, among others) without measuring a full MIC or zone.

Bactericidal testing

Minimum bactericidal concentration (MBC), also called minimum lethal concentration: dilutions showing no visible growth on the MIC series are subcultured to antibiotic-free agar; the MBC is the lowest concentration yielding ≥99.9% kill of the original inoculum. A minority of surviving “persister” cells that remain fully susceptible on retesting can confound this endpoint, as can the Eagle (paradoxical) effect, in which very high drug concentrations produce less killing than intermediate ones.

Application: MBC and the persister/tolerance concept. Continuing the Drug Y case above (MIC = 4 µg/mL), the clear wells at 4, 8, 16, and 32 µg/mL are subcultured to drug-free agar. The starting inoculum was about 4 × 10⁵ CFU/mL, so 99.9% killing requires no more than 400 surviving CFU/mL. The 4-µg/mL subculture remains too numerous to count, but the 8-µg/mL subculture yields 250 CFU/mL. Thus the MBC is 8 µg/mL and the MBC:MIC ratio is 8:4 = 2, far below the stated ≥32 ratio used to flag antimicrobial tolerance.

Automated and molecular/rapid AST

Commercial automated broth microdilution systems are the predominant testing method in most U.S. clinical laboratories. Turbidimetric or fluorometric platforms read inoculated microtiter cards or trays at intervals and produce an MIC and interpretive category within hours. Their advantages include efficient paired identification and susceptibility inoculation, integrated data-management and expert-system software, and direct interfacing with the laboratory information system. System selection weighs testing volume, technical complexity, data-management usability, and the total cost of hardware, consumables, and service against the simplicity and lower cost of disk diffusion, which remains accurate and clinically adequate for many organism-drug combinations. Any automated system must be locally verified against CLSI performance criteria, generally ≥90% essential agreement (MIC within one doubling dilution of a reference method) and ≥90% categorical agreement (matching S/I/R call). Reverification is required whenever breakpoints are revised or system software changes.

Molecular AST detects resistance genes directly by NAAT, often in a multiplex format, rather than measuring phenotypic growth inhibition. It can produce results within hours and may be performed directly from a positive blood-culture bottle or a rectal/stool screening swab. Interpretation is more complex than for molecular species identification. Resistance to one drug can arise from multiple independent genes, particularly in gram-negative organisms; detecting a gene does not confirm expression of a resistant phenotype; and specimen processing requires extraction and inhibitor removal. Molecular AST is most mature for mechanisms governed by a single dominant gene, especially mecA-mediated staphylococcal methicillin resistance and vanA/vanB-mediated enterococcal glycopeptide resistance. It is also useful for slow-growing organisms such as the Mycobacterium tuberculosis complex, for which phenotypic susceptibility results can otherwise take weeks.

Other emerging non-culture-dependent technologies include fluorescence in situ hybridization with time-lapse growth imaging, which provides identification and an MIC directly from a positive blood-culture bottle within a few hours. In MALDI-TOF-based hydrolysis assays, an antibiotic is incubated with the organism and the supernatant is analyzed by mass spectrometry for the hydrolyzed, inactive drug product. This detects functional resistance rather than only the potential encoded by a gene. Whole-genome sequencing can infer a resistance genotype from an assembled genome without prior knowledge of the resistance mechanism sought. Some studies have also used machine learning on genomic data to predict MIC directly. Cost, incomplete knowledge of novel resistance determinants, and lack of standardization currently confine WGS-based AST chiefly to epidemiologic and outbreak investigation rather than primary clinical reporting.

Mechanisms of action of the major antibiotic classes

Each major antibiotic class acts on cell-wall synthesis, the ribosome, nucleic acid synthesis, or the cell membrane, or interrupts a metabolic pathway.

Testing selection and reporting policy

Testing is warranted when an organism’s susceptibility is unpredictable. Streptococcus pyogenes pharyngitis needs no penicillin susceptibility testing because resistance has not been documented in the United States. Anaerobic susceptibility testing is indicated for isolates from brain abscess, endocarditis, osteomyelitis, joint infection, prosthetic/vascular-device infection, and bacteremia (CLSI M11); most Bacteroides group organisms are presumed penicillin-resistant without needing β-lactamase testing, but a chromogenic-cephalosporin β-lactamase test is available for other anaerobic genera. The CLSI-recommended method is agar dilution (broth microdilution is also acceptable specifically for the B. fragilis group).

Testing-menu selection is coordinated with the institution’s antimicrobial stewardship team and Pharmacy and Therapeutics Committee. Commercial automated platforms constrain which drugs can be tested together, so formulary and menu decisions must be made jointly. CLSI’s yearly M100 supplement (36th edition, 2026) groups recommended drugs by pathogen category (common nonfastidious organisms; fastidious organisms) and by testing/reporting tier: Tier 1 (routine primary testing and reporting), Tier 2 (test routinely and report selectively by cascade rule), Tier 3 (chiefly for institutions serving patients at high risk for multidrug-resistant organisms, with reporting by cascade rule), and Tier 4 (tested or reported by request or in special circumstances). Agents for urine-only isolates carry the U designation. Two drugs listed together separated by “or” predict each other’s result essentially interchangeably; drugs grouped in the same box without “or” have similar clinical efficacy but did not meet the statistical criteria for fully interchangeable reporting.

Reporting policy must account for clinical appropriateness independently of the in vitro result. Drugs that do not penetrate an infection site should not be reported for isolates from that site. Most first- and second-generation cephalosporins, macrolides, clindamycin, fluoroquinolones, tetracyclines, and oral agents are withheld from CNS-source isolates regardless of in vitro susceptibility. Drugs shown ineffective in vivo despite in vitro susceptibility are also excluded. Antimicrobial stewardship programs often use a gatekeeper model that requires infectious-disease approval before certain agents are dispensed. These programs aim to curb inappropriate use and combat resistance. Some institutions controlled outbreaks of fluoroquinolone-resistant hypervirulent Clostridioides difficile only by prohibiting fluoroquinolone use outright.

Antibiograms and stewardship

An antibiogram compiles a facility’s cumulative susceptibility data as the percentage susceptible by species and drug over a defined period. Clinicians use it to select empiric therapy pending final culture results. CLSI document M39 provides methodology guidance on terminology, data handling, statistics, and known limitations. Laboratories must decide how to count repeat isolates from one patient. Including every isolate biases the antibiogram toward patients with resistant, hard-to-treat, or long-hospitalized infections and systematically understates population susceptibility. Most guidance recommends counting only the first isolate per patient per analysis period.

General resistance mechanisms and gene transfer

Resistance genes disseminate vertically through clonal expansion under antibiotic selection pressure or horizontally through transformation, transduction, and conjugation. Transformation is the uptake of free DNA, chromosomal or plasmid. Transduction is bacteriophage-mediated transfer. Conjugation is direct cell-to-cell plasmid transfer. Conjugation can cross species boundaries; vancomycin-resistant Enterococcus faecalis has transferred resistance to Staphylococcus aureus, producing vancomycin-resistant S. aureus.

Genotypic (molecular) resistance detection and phenotypic MIC determination provide complementary information. A detected resistance gene is decisive evidence when an MIC is near a breakpoint, and molecular detection can bypass culture for speed or safety. A negative molecular result excludes only the mechanisms targeted by that assay. A resistant phenotype can still arise from an untested gene, a novel mutation, or a permeability or efflux change that the panel does not interrogate.

Staphylococcal methicillin resistance

Methicillin/oxacillin resistance in S. aureus and coagulase-negative staphylococci arises from PBP2a (also called PBP2′), an altered penicillin-binding protein encoded by mecA. The gene is carried on the mobile staphylococcal cassette chromosome mec (SCCmec) and confers resistance to essentially all β-lactams except a few newer anti-MRSA cephalosporins. Resistance can be expressed homogeneously, which is easily detected, or heterogeneously, in which a small resistant subpopulation may appear falsely susceptible by standard dilution testing. Adding 2% NaCl to the test medium improves sensitivity for heteroresistant strains during oxacillin dilution testing. Cefoxitin disk and MIC surrogate testing are performed without that salt supplement.

Alterations in PBPs, rather than β-lactamase production, similarly underlie penicillin/cephalosporin resistance in Streptococcus pneumoniae. An oxacillin disk screen is a longstanding CLSI method for non-meningitis S. pneumoniae isolates. Isolates from meningitis, and isolates that fail the screen, require full MIC testing because meningitis breakpoints are stricter and because the screen cannot reliably separate susceptible, intermediate, and resistant isolates within the failed-screen range. Exact millimeter cutoffs remain in the current M100-Ed36 S. pneumoniae table and are not quoted here.

Streptococcal and enterococcal resistance

Group A Streptococcus (S. pyogenes) remains uniformly penicillin-susceptible; testing is unnecessary except when penicillin allergy raises the question of a macrolide/lincosamide alternative, since erythromycin resistance is a recognized emerging problem in β-hemolytic streptococci generally.

Isolates of staphylococci, β-hemolytic streptococci, or pneumococci that test erythromycin-resistant but clindamycin-susceptible may carry an inducibly expressed erm gene (macrolide-lincosamide-streptogramin B, MLSB resistance). Clindamycin appears active in vitro because it is a weak inducer of the erm methylase. Exposure to the stronger inducer erythromycin can activate resistance during therapy and cause clinical failure. The D-zone approximation test places clindamycin and erythromycin disks close together on an inoculated plate. Flattening of the clindamycin zone into a “D” shape on the side facing the erythromycin disk indicates induction, and the isolate is reported clindamycin-resistant despite an apparently susceptible standalone result. This test is specifically recommended when screening penicillin-allergic, group B Streptococcus-colonized pregnant women for intrapartum prophylaxis options.

Enterococcus:

• Penicillin/ampicillin. Penicillin susceptibility predicts ampicillin susceptibility; the reverse prediction does not hold. Most resistance, chiefly in E. faecium, is due to low-affinity PBPs and is not detected by β-lactamase testing. A minority of isolates produce β-lactamase, which dilution MIC testing also cannot detect. The nitrocefin direct β-lactamase test is therefore recommended specifically for isolates from blood or CSF.
• High-level aminoglycoside screening. Aminoglycosides are clinically inactive against enterococci when used alone and are tested only for potential synergy with a cell-wall-active agent such as penicillin, ampicillin, or a glycopeptide. Screening uses 500 μg/mL gentamicin or 1,000 μg/mL streptomycin by broth dilution. Synergy requires the aminoglycoside to reach the cytoplasm after the cell-wall agent has compromised the envelope. Growth at these high concentrations predicts the absence of useful synergy.
• Glycopeptide (vancomycin) resistance. Intrinsic, low-level, non-transferable resistance occurs in E. gallinarum and E. casseliflavus/E. flavescens and has no infection-control implication. Acquired vancomycin resistance occurs chiefly in E. faecium and E. faecalis. Altered pentapeptide peptidoglycan precursors have roughly 1,000-fold lower vancomycin-binding affinity. They are encoded by vanA, which confers high-level resistance to vancomycin and teicoplanin, or vanB, which confers lower-level vancomycin resistance with retained teicoplanin susceptibility. Screening uses vancomycin-containing (6 μg/mL) brain-heart infusion agar or a chromogenic VRE-selective medium. Read the medium only after a full 24-hour incubation because earlier reads risk false-negative results. Molecular vanA/vanB PCR confirms the phenotype and, in surveillance settings, can screen stool or rectal specimens directly for VRE carriage.

Vancomycin-intermediate S. aureus (VISA, MIC 4-8 μg/mL) results from increased cell-wall thickness and altered vancomycin-binding sites. Vancomycin-resistant S. aureus (VRSA, MIC ≥16 μg/mL) results from acquisition of the enterococcal vanA gene by conjugation. Both are almost always also MRSA, and vancomycin is a default empiric MRSA agent. These S. aureus vancomycin MIC categories remain current in M100-Ed36. Disk diffusion is not a reliable screen, vancomycin-screening agar misses isolates with MICs at the lower end of the intermediate range, and automated systems vary in sensitivity. These limitations should be considered when a patient fails vancomycin therapy despite a susceptible result. MIC creep within the susceptible range has also been described.^1,2

Gram-negative beta-lactamases: ESBL, AmpC, carbapenemase

Hundreds of β-lactamase enzymes have been described in gram-negative organisms. The Ambler scheme classifies them structurally into molecular classes A-D based on the catalytic-site amino acid motif. The Bush-Jacoby scheme classifies them functionally by substrate and inhibitor profile.

ESBL confirmation (when performed) uses the susceptibility of the ESBL enzyme to β-lactamase inhibitors. A decreasing MIC or growing zone for a third-generation cephalosporin (cefotaxime or ceftazidime) tested with clavulanate, compared with the cephalosporin alone, confirms ESBL production. Since lowered CLSI cephalosporin breakpoints were adopted, routine ESBL confirmatory testing is no longer strictly required for reporting because the MIC-based category alone is considered predictive of clinical response. The phenotypic test remains useful for infection-control and epidemiologic purposes. Published evaluations have shown that a meaningful fraction of ESBL producers still test susceptible or intermediate to some cephalosporins/aztreonam under the newer breakpoints.

Because chromosomally encoded, inducible AmpC occurs in the organisms listed above, testing for it in those genera adds little: consider them intrinsically resistant to broad-spectrum penicillins, extended-spectrum cephalosporins (except cefepime at conventional inoculum), monobactams, and cephamycins regardless of an individual in vitro result. Where plasmid-mediated AmpC is suspected in an organism that does not normally carry it, phenotypic screens include cefoxitin resistance combined with a positive ESBL screen but negative ESBL confirmatory test (suggesting AmpC rather than ESBL), and inhibitor-based confirmatory tests using boronic acid or cloxacillin; cefepime is a relatively stable poor substrate for AmpC. Paired cefoxitin disks with and without boronic acid or cloxacillin show an increased zone around the inhibitor-containing disk when AmpC is present. Phenotypic testing cannot reliably separate chromosomal from plasmid-mediated AmpC; that distinction requires molecular testing.

Carbapenemase-producing organisms (CPOs), including carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE), are an infection-control priority.

Neisseria and Haemophilus resistance

Haemophilus influenzae β-lactamase production is reliably screened by the nitrocefin test. A smaller fraction of strains are β-lactamase-negative, ampicillin-resistant (BLNAR). These strains carry altered PBPs, are not detected by the β-lactamase test, and require full susceptibility testing. N. meningitidis penicillin resistance is likewise primarily PBP-mediated, although occasional β-lactamase producers occur. N. gonorrhoeae has accumulated altered-PBP and β-lactamase-mediated penicillin resistance, as well as rising cephalosporin and fluoroquinolone resistance. Culture with susceptibility testing is recommended in cases of apparent treatment failure.

Mycobacterium tuberculosis molecular resistance detection and susceptibility testing are covered in mycobacteriology.

Colonization screening: specimen sources and methods

Infection-control colonization surveillance detects carriage of an organism capable of transmission to other patients or later invasive disease, independent of current infection status. Specimen source is organism-specific:

Chromogenic culture is inexpensive, detects any resistant organism present regardless of mechanism, and confirms viable colonization, but requires overnight incubation. NAAT returns a same-day result and identifies the specific resistance gene, but a gene-targeted panel can miss an organism carrying a mechanism outside its design. Many infection-control programs use chromogenic culture for routine surveillance and reserve NAAT for high-acuity situations (outbreak response, ICU/transplant admission) where turnaround time drives the choice.