Common Misinterpretations of EC50

Why EC50 Is Easy to Misread

EC50 is often described informally as “the concentration that gives 50% of the maximum response.” That wording can be ambiguous. In a baseline-relative Emax model, the maximum drug-related effect is the increase from baseline to the model’s upper plateau. EC50 marks the midpoint of that increase; it does not necessarily produce half the numerical value of the total response.

The distinction matters whenever the baseline is not zero on the response scale. A response of 60 units might be halfway through an effect that rises from 20 to 100 units, but it is not half of the absolute plateau of 100. These are different targets, and they generally occur at different concentrations.

Here \(C\) is concentration, and the model assumes \(C\geq 0\), \(EC_{50}>0\), and a positive \(E_{\max}\) for an increasing response. It describes a direct, instantaneous concentration-effect relationship. Its upper asymptote is a modelled limit as concentration increases, not necessarily a response attained or observed in the data.

What Happens at EC50?

Substitute \(C=EC_{50}\) into the model. The concentration-dependent fraction is one-half, so the response is baseline plus half the drug-related maximum:

Thus “50%” refers to the fraction of the modelled drug effect, \(R(C)-E_0\), relative to its maximum, \(E_{\max}\). If the response scale has a nonzero baseline, this midpoint need not be half of the absolute upper asymptote. The distinction disappears when the response has been defined or transformed so that \(E_0=0\), provided the model and response scale otherwise remain the same.

Finding the Concentration for Half the Absolute Plateau

To find the concentration at which response is half the absolute upper asymptote, first set the response equal to \((E_0+E_{\max})/2\), then solve the model for \(C\). This is a different target from setting the drug-related increase to \(E_{\max}/2\).

Subtracting \(E_0\) from both sides gives the required drug-related increase. Rearranging then gives the concentration, when the target is attainable at a positive concentration:

For the common case of positive \(E_0\) and \(E_{\max}\), the half-absolute-plateau target lies above baseline only if \(E_{\max}>E_0\). If \(E_{\max}=E_0\), it is exactly the baseline response, reached at zero concentration. If \(E_{\max}<E_0\), half the absolute plateau is below baseline, so an increasing model starting at \(E_0\) does not reach that target. The algebraic expression should not be reported as a positive concentration in either of those cases.

When Do the Two Concentrations Coincide?

For the increasing Emax model above, EC50 gives a response of \(E_0+E_{\max}/2\), while half the absolute plateau is \((E_0+E_{\max})/2\). Setting these targets equal shows exactly when they coincide:

So, for this model on its stated response scale, the two meanings coincide when baseline is zero. They do not become equivalent merely because the same parameter name is used, or because concentrations have been scaled. If response is expressed as change from baseline, the shifted response has zero baseline by construction; the resulting half-effect interpretation is then relative to that shifted scale.

The same distinction carries over to the sigmoid Emax model, in which \(h\) controls the shape of the concentration-effect curve:

At \(C=EC_{50}\), the fraction is still one-half, so the response is \(E_0+E_{\max}/2\). For a positive baseline with \(E_{\max}>E_0\), solving for half the absolute plateau instead gives \(C=EC_{50}[(E_{\max}-E_0)/(E_{\max}+E_0)]^{1/h}\). The baseline determines the difference in target, while \(h\) changes how that difference maps to concentration.

Interpretation, Diagnostics, and Alternatives

When reporting EC50, state the response scale and the reference point for “half.” A useful interpretation is: “At 4 mg/L, the model predicts half of the maximum drug-related increase above baseline.” Avoid saying “the response is 50%” unless the response was defined on a zero-based scale and the statement is genuinely about that scale. EC50 has concentration units; the response midpoint has response units.

A simple diagnostic is to calculate both \(R(EC_{50})\) and the upper asymptote from the fitted parameterization. If the baseline is nonzero, compare \(R(EC_{50})\) with \(E_0+E_{\max}/2\) and with \((E_0+E_{\max})/2\). This check can reveal a verbal interpretation that does not match the model. Also check that “maximum” refers to the modelled asymptote, not the largest observed response, which may be lower because the observed concentrations do not reach the plateau.

A common mistake is to read the subscript “50” as automatically meaning 50% of the total response. In the baseline-relative model, that is wrong: EC50 is defined by half of \(E_{\max}\), the increment above baseline. The correction is to identify the reference level first, then solve for the concentration corresponding to the response target of interest.

Another source of confusion is a different parameterization in which a symbol called \(E_{\max}\) denotes the upper response itself rather than an increment above \(E_0\). Parameter labels alone do not settle the meaning. Write the equation, identify its baseline and upper asymptote, and determine what response is predicted at the parameter called EC50. For inhibitory responses or other response scales, the direction and reference level also need explicit interpretation; do not transfer an increasing-response statement without checking the model.

The algebra gives a model-based target, not evidence that the target is clinically meaningful, well estimated, or observed in the data. If the response scale has a nonzero baseline, changing to a change-from-baseline scale makes the relative-effect midpoint numerically equal to half of that shifted response plateau, but it does not change the underlying curve’s predictions. The two target definitions should remain explicit when comparing parameterizations or communicating results.