Applying Loewe Additivity to Dose-Equivalent Agents

Why Dose Equivalence Provides an Additivity Reference

For two drugs that can produce a comparable effect, a useful question is whether a combination reaches that effect with less total dose than expected if one drug could substitute for the other. Loewe additivity formalizes this dose-substitution idea. It is especially interpretable for agents with similar or overlapping actions, but it is not a universal definition of “no interaction”: the conclusion depends on whether dose equivalence is a reasonable description of the agents and endpoint.

The calculation is made at a particular effect level. First find how much of drug A alone and drug B alone would produce that effect. Then compare the combination doses with those monotherapy requirements. A combination index below one indicates that the combination uses less than one additive equivalent dose; a value above one indicates more. The index does not, by itself, establish a molecular mechanism.

Definition: Let $D_A(E)$ and $D_B(E)$ denote the monotherapy doses of drugs A and B, respectively, that produce effect $E$. If a combination containing doses $a$ and $b$ produces the same effect $E$, its Loewe combination index is $$ CI(E)=\frac{a}{D_A(E)}+\frac{b}{D_B(E)}. $$ The Loewe additive reference is $CI(E)=1$. A value $CI(E)<1$ is described as synergistic relative to this reference, and $CI(E)>1$ as antagonistic relative to it.

Each ratio is a fraction of the dose of one drug that would be required alone to produce $E$. The doses need not have the same physical units: each dose is divided by a monotherapy dose in the same units, so the index is dimensionless. The combination must actually produce the effect being assessed; comparing doses at different effects does not define this index.

Assumptions and a Short Derivation

Loewe additivity treats the two drugs as substitutable contributors to a common effect. At a fixed effect $E$, one full monotherapy-equivalent dose of either drug reaches the effect alone. If part of that requirement is supplied by drug A and the remainder by drug B, additivity assigns the combination a total of one equivalent dose.

To express this idea, a dose $a$ of A supplies the fraction $a/D_A(E)$ of A’s dose requirement at that effect, while dose $b$ of B supplies $b/D_B(E)$. Under dose substitution, those fractions add. The additive boundary is therefore obtained by setting their sum equal to one:

\frac{a}{D_A(E)}+\frac{b}{D_B(E)}=1.

This is the fixed-effect Loewe isobole. It is a straight line in the plot of dose $a$ against dose $b$, with intercepts $D_A(E)$ and $D_B(E)$. The line is not a claim that the biological response itself is linear in dose; it follows from the dose-equivalence assumption at the selected effect.

Interpretation requires that the monotherapy dose–response relationships are sufficiently well characterized and that both drugs can produce the selected effect. The effect scale and endpoint must be comparable across monotherapy and combination measurements. Dose–response curves should be monotone over the relevant range so that a target effect maps to a meaningful dose. If a drug cannot reach $E$, its dose requirement is undefined, and this form of the index cannot be used for that target. Finally, the dose-substitution premise must be scientifically defensible; two drugs acting through unrelated processes may not be well represented by this reference.

Worked Example: Calculating a Combination Index From Equivalent Doses

Suppose hypothetical monotherapy experiments show that 8 mg of A or 20 mg of B alone produces 60% inhibition of a cell-growth endpoint. A combination of 3 mg A and 10 mg B also produces 60% inhibition. Calculate the Loewe index and interpret it.

At the specified effect, $D_A(E)=8$ mg and $D_B(E)=20$ mg. The observed combination doses are $a=3$ mg and $b=10$ mg, so

CI(E)=\frac{3\ \mathrm{mg}}{8\ \mathrm{mg}}+ \frac{10\ \mathrm{mg}}{20\ \mathrm{mg}} =0.375+0.500=0.875.

The units cancel in each ratio. Because $CI=0.875<1$, this combination uses 0.875 additive-equivalent dose units to produce 60% inhibition. It is synergistic relative to Loewe additivity at this effect level. This result does not establish why the combination reaches the effect with fewer equivalent doses.

Reading and Constructing Isoboles

An isobole displays all dose pairs predicted to yield a selected effect under the reference model. For effect $E$, the intercept on the A axis is $D_A(E)$, and the intercept on the B axis is $D_B(E)$. A point on the line has $CI=1$. A measured combination point below and to the left of the line has $CI<1$; a point above and to the right has $CI>1$. This geometric description assumes the axes are the doses of A and B and that the same effect level is being compared throughout.

The procedure can be applied separately at several effect levels. Each level has its own monotherapy intercepts and hence its own isobole. Comparing a combination at one inhibition level with monotherapy intercepts taken from another level can produce a meaningless apparent departure.

Worked Example: Locating a Combination on a Fixed-Effect Isobole

In a hypothetical assay, 6 mg of A alone and 12 mg of B alone each produce 50% inhibition. A combination of 2 mg A and 5 mg B also produces 50% inhibition. Construct the additive isobole and locate the combination relative to it.

The additive isobole at 50% inhibition is

\frac{a}{6\ \mathrm{mg}}+\frac{b}{12\ \mathrm{mg}}=1.

Its axis intercepts are $(6\ \mathrm{mg},0)$ and $(0,12\ \mathrm{mg})$. For the observed combination,

CI=\frac{2}{6}+\frac{5}{12} =0.3333+0.4167=0.7500.

The combination point has $CI=0.75$, so it lies below and to the left of the additive line in dose space. Under the Loewe reference, it is synergistic for 50% inhibition. A point such as $(2\ \mathrm{mg},8\ \mathrm{mg})$ would lie on the line because $2/6+8/12=1$; it represents an additive combination at this effect.

Obtaining Equivalent Doses From Dose–Response Curves

The monotherapy-equivalent doses may come from measured dose–response data or from a fitted model. For example, consider an inhibitory effect described as a fraction of the maximum scale by a Hill function:

E(D)=E_{\max}\frac{D^h}{ED_{50}^{\,h}+D^h},

where $E_{\max}$ is the maximum effect, $ED_{50}$ is the dose producing half of that maximum, and $h>0$ is the Hill coefficient. For a target $0<E<E_{\max}$, solving for dose gives

D(E)=ED_{50}\left(\frac{E}{E_{\max}-E}\right)^{1/h}.

This inversion makes the domain restriction explicit: a target at or above $E_{\max}$ does not yield a finite dose from this model. If drugs have different maximum effects, the target must be achievable for both before their dose requirements can be compared.

Worked Example: Inverting Hill Curves Before Calculating the Index

Consider hypothetical monotherapy curves for an inhibition endpoint. Drug A has $E_{\max}=0.90$, $ED_{50}=4$ mg, and $h=1$. Drug B has $E_{\max}=1.00$, $ED_{50}=8$ mg, and $h=2$. A measured combination of 2 mg A and 4 mg B produces inhibition $E=0.60$. Find the monotherapy-equivalent doses and the index.

The target $0.60$ is below both maximum effects, so it is within both curves’ ranges. For A,

D_A(0.60)=4\left(\frac{0.60}{0.90-0.60}\right)^{1/1} =4(2)=8\ \mathrm{mg}.

For B,

D_B(0.60)=8\left(\frac{0.60}{1.00-0.60}\right)^{1/2} =8\sqrt{1.5} \approx 9.798\ \mathrm{mg}.

Using the combination doses that produced the same 0.60 effect,

CI(0.60)=\frac{2}{8}+\frac{4}{9.798} \approx 0.2500+0.4082=0.6582.

The point estimate is below one, indicating a departure toward synergy relative to Loewe additivity at 60% inhibition. The calculation depends on the fitted curves: uncertainty or misspecification in their maxima, slopes, or dose estimates can change the index.

Diagnostics, Alternatives, and Limits

Before interpreting a calculated index, check the inputs and the shape of the data. Useful checks include:

Matched effect: confirm that the combination and both monotherapy doses correspond to the same endpoint, time point, and effect level.
Curve support: check that the monotherapy data cover the target effect rather than requiring a long extrapolation. A target near an estimated maximum can produce a highly unstable dose inverse.
Monotonicity and plateaus: inspect whether the response is monotone over the relevant range and whether a plateau or nonmonotone region makes a unique dose requirement doubtful.
Reproducibility: assess whether small changes in the combination response or fitted monotherapy curves cause a large change in $CI$. A small numerical departure from one should not be treated as decisive when the estimates are imprecise.
Comparable exposure: verify that nominal doses meaningfully represent comparable exposures in the combination and monotherapy settings. A pharmacokinetic change can alter exposure without establishing a pharmacodynamic interaction.

A frequent mistake is to calculate $a/D_A(E)+b/D_B(E)$ using monotherapy doses from different effect levels, then label the result synergy or antagonism. That sum has the intended interpretation only when the denominator doses both produce the same target effect as the combination. Another mistake is to treat $CI<1$ as proof of a specific biological interaction. The correction is to state the effect level, describe the result as relative to Loewe additivity, and distinguish the model-based departure from a demonstrated mechanism.

Loewe additivity differs from Bliss independence, derived in the earlier material on Bliss independence. Bliss uses a reference based on independent probabilities of effects, whereas Loewe uses dose substitution at a common effect. When the drugs are not meaningfully dose-equivalent, the Loewe assumption may be less persuasive; when they can substitute for one another, it can provide a direct and interpretable reference. The two references can therefore give different classifications without either arithmetic calculation being incorrect. Choosing which reference is scientifically appropriate depends on the agents, endpoint, and question.

The index is also effect-specific: a pair of drugs may have $CI<1$ at one level and $CI>1$ at another. A single number should not be taken as a summary of every dose pair or every endpoint. Moreover, the point estimate alone does not quantify uncertainty. Conclusions should be qualified when monotherapy curve estimates or combination responses are noisy, or when the combination lies close to the additive isobole.

Key takeaway: A Loewe combination index compares combination doses with the monotherapy doses required for the same effect. Its interpretation as additive, synergistic, or antagonistic is conditional on dose equivalence, valid dose–response estimates, and a matched effect—not a universal statement about the mechanism of a drug pair.

Check Your Understanding

Work through each item, then compare with the solutions below.

At a target effect, $D_A(E)=10$ mg and $D_B(E)=5$ mg. A combination uses 4 mg A and 3 mg B. Calculate $CI$ and interpret it relative to Loewe additivity.
For the same target effect, identify the additive B dose paired with 3 mg A if $D_A(E)=9$ mg and $D_B(E)=18$ mg.
A monotherapy curve has $E_{\max}=0.80$. Can its dose requirement be obtained from that curve for a target effect $E=0.85$? Explain.
Why is it invalid to use $D_A(40\%\text{ effect})$ and $D_B(60\%\text{ effect})$ together in the index for a combination producing 60% effect?

Solutions. (1) $CI=4/10+3/5=0.40+0.60=1.00$, so the combination is on the Loewe additive reference at that effect. (2) Solve $3/9+b/18=1$, giving $b/18=2/3$ and $b=12$ mg. (3) No. The target exceeds the curve’s maximum effect, so there is no finite monotherapy dose requirement under that model. (4) The denominator doses must both represent the same target effect as the combination. Mixing effect levels makes the dose fractions incomparable and the resulting sum is not a Loewe index for 60% effect.