IC50 and EC50 Definitions

In pharmacology, the principles of IC50 and EC50 are central. The EC50 is the drug concentration that produces a half-maximal response. The IC50 is the concentration of an inhibitor  at which the response (or binding) is decreased by half.

Although the concept is simple, there are a few challenges when try to fit data to find these values.

We are considering IC50 in this post (I for inhibition, for downward sloping dose-response curves). All the concepts can be used to stimulatory curves and EC50 (E for effective) also.

IC50 – The perfect situation 

This figure shows a perfect situation:

Measurements taken with controls are indicated by the green symbols. Since the ones on the left (Blank) do not have an inhibitor, describe “100% “. The ones on the right are in the presence of a standard inhibitor at its maximum concentration, so describe “0 percent “. The experimental dose-response curve’s data (red dots) span the entire range between the two control values.

When fitting this curve, you must determine how to fit the curve’s top plateau. There are three options available to you:

• Ignore the Blank control values and just fit the data.
• Set the parameter Top to be a constant value equal to the mean of the blanks, and average the Blank control values.
• As if the blank values were part of the dose-response curve, enter them. Simply put, start with a low dose, such as 10-10 or 10-11. You can’t enter zero, because zero is not defined on a log scale.

Since the data shape a full dose-response curve with a simple top plateau that is indistinguishable from the null, the results would be quite similar with either of these approaches. The third approach is my favorite because it analyses all of the data, but it is not a strongly preferred.

Similarly, dealing with the bottom plateau can be done in three ways: Fit only the data, with Bottom set to the average of the NS controls and the NS controls treated as if they were a very high concentration of inhibitor in the fit.

That is a perfect scenario. There’s no misunderstanding on what IC50 stands for.

A case where IC50 is described in two ways

There are two ways to describe IC50 in this case.

The inhibition curve plateaus well above the control values (NS) established by a high concentration of a standard drug in this illustration. As a result, different meanings of IC50 emerge.

Clearly, such a curve cannot be summarised by a single value. You’d need at least two values: one to measure the drug’s efficacy in the centre of the curve, and another to quantify the drug’s maximum effect at the bottom.

The graph above depicts two IC50 concepts.

The relative IC50 definition is  the most common, and the adjective relative is generally omitted. It is the concentration needed to bring the curve halfway between the top and bottom plateaus. With this concept of IC50, the NS values are completely ignored. This is the term that is used in traditional pharmacological study of agonist and antagonist interactions. Estimated Kd values can also be derived from the IC50 value described this way, with proper consideration of the biological system and concentrations of interacting ligands (not so for the “so-called absolute IC50” mentioned below).

The absolute IC50 is the concentration that produces a result halfway between the Blank and NS values. The horizontal dotted lines indicate how 100 percent and 0 percent are described, which then defines 50 percent. This is a word that isn’t widely used. The authors of the International Union of Pharmacology Committee on Receptor Nomenclature believe that this value does not accurately represent a drug’s potency. (1) think that the concept of absolute IC50 (and that term) is not useful (R. Neubig, personal communication). I agree.

Drugs that delay cell growth are measured using this concept (but not the expression “absolute IC50”). For what we refer to as the total IC50, the abbreviation GI50 is used. The Environmental Protection Agency (EPA) uses it to assess endocrine disrupters (Appendix A). The absolute IC50 is referred to as IC50, and the relative IC50 is referred to as EC50. They don’t use the terms relative and absolute.

Incomplete dose-response curves

Attempting to calculate IC50 by fitting a curve to the data in the graph above would be futile. A curve fitting software may or may not be able to match the data to a dose-response curve. However, if the curve matches, the IC50 value is likely to be irrelevant, with a very large confidence interval. The data does not shape a top plateau (which would define 100) or a bottom plateau (which would define 0). If the data hasn’t specified 100 or 0, then 50, like the IC50, is also undefined.

The curve can be easily fit if you already have control values that describe 100 and 0. Fitting a dose response curve when constraining the Top plateau to be a constant value equal to the mean of the Blanks values and the Bottom plateau to be equal to the mean of the NS values yielded the curve below.

The IC50 value fits this way only if you believe that higher concentrations of the inhibitor would ultimately inhibit down to the NS values. That is a hypothesis that cannot be verified with the available data.

In this case, the distinction between relative and absolute IC50 is irrelevant. The IC50 must be specified in relation to the NS control values because the data do not describe a bottom plateau.

Standardized Data Fitting – IC50

As you can see from the examples above, normalising the data to run from 100 percent to 0 percent is not needed. Curves may be fitted using data in their natural units. It’s a common misconception that normalising data before fitting dose-response curves is needed.

If you want to normalize your data, it is critical that you carefully consider (and document in the methods section of your paper) how 100% and 0% are represented. You should employ one of three strategies:

• Controls from the outside (Blank and NS in the figures above). Since these values are so significant, consider using more replicates for these controls than for the rest of the experiment.
• From very low and very high concentrations of the test substance.
• From the nonlinear regression plateaus. Fit the curve first, then normalise the data using the best-fit values of the Top and Bottom plateaus.

If you’re fitting normalised results, you expect GraphPad Prism to probably force the curve to go from 100 to 0. If you don’t tell it, it won’t know how to do it. Make sure you’re not making the common mistake of normalising the data without constraining the curve to go from 100 to 0. There are two ways to constrain the curve:

• Select a normalised model to fit to. Prism’s built-in normalised models are often between 0 and 100.
• Set Bottom to a constant value of 0.0 and Top to a constant value of 100.0 in a more general model, using the  Constrain tab.

Conclusion for EC50 and IC50

Unless you define the values clearly for 100 percent and 0 percent, the definition of IC50 (or EC50) is a little unclear.

Reference

1. R. R. Neubig et al. International Union of Pharmacology Committee on Receptor Nomenclature and Drug Classification. XXXVIII. Update on terms and symbols in quantitative pharmacology. Pharmacol Rev (2003) vol. 55 (4) pp. 597-606