In drug development, there are certain terms that people throw around that signifies that they know something about a field they’re not an expert in. These are shorthands for asking, is that compound any good?

“What’s the p-value?” (buzzword for clinical data)

“Does it meet Lipinski’s Rule of Five?” (buzzword for medicinal chemistry)

“What’s the IC50?” (buzzword for in vitro physiochemical characterization)

Any term that serves as a surrogate to simplify a wide range of data into one number can be dangerous, but IC50 in particular is insidious.

IC50 is a measurement of how much drug it takes to inhibit a biological or chemical process by half. For example, if you have an inhibitor of a kinase (an enzyme that adds phosphate groups to proteins, often a target for cancer and other diseases), IC50 might be the concentration of your drug that blocks the kinase activity by 50%.

Often, people compare across drug candidates by talking about IC50. They will say things like “Wow, that compound has IC50 of 2 nanomolar. That’s a really tight binder, it is a good drug candidate.”

Apart from the fact that IC50 are a logarithmic value, there is a terrible pitfall in using IC50. The IC50 varies wildly depending on the conditions under which it’s measured. Specifically, the concentration of the natural ligand has a profound influence on the IC50 value. For example, in the case of the kinase, if you raise the concentration the protein that’s being phosphorylated by the kinase, then IC50 will go up. You can make any molecule have nanomolar IC50 if you use very low concentrations of the natural ligand. Here is a good slide presentation on why.

It makes sense, right? The drug is competing against the natural ligand for the binding site, so of course IC50 will be higher if the competing ligand concentration is higher.

So, IC50 means nothing without the context. It is not the same as affinity, or Ki, or any one of a number of properties we might associate with how good a molecule is. In fact, it is a often-misleading term that we should avoid using.