The minimum inhibitory concentration is the lowest concentration of an antibacterial that prevents visible growth of an organism after a defined incubation. It is the number every antibacterial program lives by, and it hides more convention than most people expect.
The definition, precisely
A MIC is defined against a specific organism, a specific method, a specific inoculum and a specific incubation time. Change any of those and the number moves. The reference method is broth microdilution, standardised by CLSI and by EUCAST, with a starting inoculum of roughly 5 x 105 colony forming units per millilitre and incubation of 16 to 20 hours.
The result is reported in µg/mL, which is numerically identical to mg/L. A MIC of 2 µg/mL means growth was visible at 1 and not at 2.
Why the value is always a power of two
The plate is prepared as a doubling dilution series: 0.03, 0.06, 0.12, 0.25, 0.5, 1, 2, 4, 8 and so on. The MIC is whichever of those wells is the first clear one, so the measurement has the resolution of one doubling dilution and no more. Two competent labs testing the same compound and strain routinely land one step apart, and that is considered agreement rather than error.
This is why Antibacterial predicts a band such as 0.5-2 rather than a single decimal.
A predicted 0.73 µg/mL would be false precision applied to a measurement that never had that precision.
How it is measured
- Broth microdilution. The reference. A 96 well plate, a doubling dilution series in each row, a standardised inoculum, read by eye or by optical density.
- Agar dilution. Antibacterial incorporated into agar plates at each concentration, organisms spotted on top. Laborious but good for testing many organisms against a few concentrations.
- Gradient diffusion strips. A strip carrying a concentration gradient laid on an inoculated plate. Fast and convenient, and it can read between the doubling steps, which is convenient rather than more accurate.
- Disk diffusion. Not a MIC at all. It measures a zone of inhibition in millimetres, which is then interpreted against a species specific table.
Quality control strains with known expected ranges are run alongside, which is how a lab knows the plate itself is behaving. Details of the bench procedure are in the blog post on how MIC is measured.
MIC50, MIC90 and breakpoints
Three related numbers are easily confused:
| Term | What it is | What it is for |
|---|---|---|
| MIC | The value for one compound against one isolate | Bench result |
| MIC50 | The concentration inhibiting half of a collection of isolates | Describing a population |
| MIC90 | The concentration inhibiting ninety percent of that collection | The number quoted in surveillance papers |
| Breakpoint | A regulatory threshold separating susceptible from resistant | Clinical interpretation, set by CLSI or EUCAST |
A breakpoint is not a property of the compound. It is a decision that takes the MIC distribution, the achievable drug exposure in a patient and the clinical outcome data into account. That is why the same MIC can be susceptible for one species and resistant for another.
Where MIC data comes from, and why it is hard to use
Published MIC values are scattered across primary literature, surveillance reports and public databases such as ChEMBL. Assembling a picture for one compound means reconciling different methods, different strain collections and different reporting conventions. It is entirely doable and it takes an afternoon per compound, which is the specific cost a screening layer removes.
Predicting a MIC band before the plate
A predicted band is not a measurement and it is not a substitute for one. What it is good for is ordering: deciding which twelve of four hundred analogs justify plate time this month, and knowing in advance which organism is likely to be the one that kills the series. On Gram-negatives that will usually be uptake or efflux rather than affinity, for the reasons set out under Gram-negative bacteria.
Run a compound you already know well through the screen below and compare the band against the literature you already trust. That is the fastest way to calibrate how much weight to put on it.