Written and maintained by CASRAI Editorial Board
Last updated
The Chou-Talalay method is the standard framework pharmacology researchers use to determine whether two drugs act synergistically, additively, or antagonistically when combined. Rather than relying on a visual impression of an isobologram or a qualitative call of “more than additive,” it produces a single quantitative synergy metric — the Combination Index (CI) — derived from each drug’s individual dose-response behavior. This guide covers the method’s foundation in the median-effect equation, how the CI is calculated and interpreted, the experimental design it requires, and a genuine methodological caution about over-reading the CI number in isolation.
The Median-Effect Equation: The Method’s Foundation
Chou and Talalay built their method on the median-effect equation, a general dose-effect relationship derived from the mass-action law:
fa / fu = (D / Dm)m
where fa is the fraction of the system (cells, enzyme activity, or another measured endpoint) affected at dose D, fu (= 1 − fa) is the unaffected fraction, Dm is the median-effect dose — the dose producing a 50% effect, equivalent to an IC50 or ED50 — and m is a shape parameter describing the sigmoidicity of the dose-effect curve. This is the same generalized, Hill-type dose-response relationship covered in CASRAI’s guide to the Hill coefficient in dose-response and binding curves: m plays the same curve-steepness role that nH plays in the Hill equation, and Chou’s own derivation shows the median-effect equation reduces to, or subsumes, the Michaelis-Menten, Hill, and Henderson-Hasselbalch equations as special cases. Dm itself is conceptually the same “concentration for half-maximal effect” quantity discussed in CASRAI’s Cheng-Prusoff equation guide, though Cheng-Prusoff converts an assay-dependent IC50 into an intrinsic Ki, while the median-effect equation instead uses Dm to characterize the shape of a single drug’s own dose-response curve as a prerequisite for combination analysis.
In practice, each drug tested alone is fit to a linearized form of this equation (plotting log[fa/fu] against log[D]) to extract its own Dm and m. Those per-drug parameters are what the Combination Index calculation below is built from — the median-effect fit is not an optional preliminary step, it is the quantitative foundation the whole method depends on.
The Combination Index (CI): A Quantitative Synergy Metric
The Combination Index is the method’s key output, and the reason it is preferred over a purely visual isobologram reading: it turns “does this combination look synergistic” into a number that can be compared across dose pairs, effect levels, and even across different experiments.
For a two-drug combination, CI at a given effect level is calculated from the doses of drug 1 and drug 2 that, combined, produce that effect (D1 and D2), divided by the doses of each drug alone that would produce the same effect on its own ((Dx)1 and (Dx)2, obtained from each drug’s median-effect equation):
CI = D1/(Dx)1 + D2/(Dx)2 (plus a third cross-term for drugs assumed to act through mutually non-exclusive mechanisms)
The interpretation is a direct, quantitative reading:
- CI < 1 — synergism: the combination produces the target effect at a lower combined dose than either drug’s individual dose-response would predict.
- CI = 1 — additivity: the combination’s effect is exactly what you’d expect from simply summing each drug’s independent contribution.
- CI > 1 — antagonism: the combination requires more drug, not less, to reach the same effect than the individual dose-response curves predict.
Because CI is computed from real dose-response data rather than assigned by eye, it supports graded language (e.g., a CI of 0.3 reflects a much stronger synergistic interaction than a CI of 0.85, even though both are “less than 1”) — researchers typically report CI across a range of effect levels (commonly fa = 0.25, 0.5, 0.75, 0.9) rather than a single value, since the strength and even the direction of the interaction can shift across the dose-effect curve.
Experimental Design: Fixed-Ratio Combinations and the Isobologram
A Chou-Talalay analysis requires three sets of dose-response data, generated in the same experiment under matched conditions:
- Drug A alone, tested across a dose range (typically 5-8 concentrations) spanning well below and above its IC50/ED50.
- Drug B alone, tested across an equivalent range spanning its own IC50/ED50.
- Drug A + Drug B in combination, tested at a fixed dose ratio (commonly set at, or near, the ratio of their individual IC50/ED50 values, so each drug contributes comparably to the combined effect) across a parallel dilution series.
All three dose-response curves are fit to the median-effect equation to obtain Dm and m for each. The combination data are then commonly visualized as an isobologram: a plot with drug A’s dose on one axis and drug B’s dose on the other, where a straight line connects the two single-agent doses that independently produce a chosen effect level (e.g., 50% inhibition). That line represents pure additivity. A combination dose pair that falls below the line uses less of both drugs than additivity would require (synergism); a pair falling on the line is additive; a pair falling above the line requires more of both drugs (antagonism). The CI calculation is the numerical counterpart of this same geometric picture, which is why the two are typically reported together rather than as alternatives.
A Methodological Caution: Don’t Read the CI Number in Isolation
The CI is a genuinely useful quantitative metric, but it is not a self-certifying one. Several factors can shift a reported CI without any change in the underlying biology:
- Dose range sensitivity. The median-effect fit that CI depends on is only reliable across the dose range where fa is neither near 0 nor near 1 — a dose range that undersamples the steep part of either drug’s curve produces a poorly constrained Dm/m estimate, which then propagates into an unreliable CI.
- The mutual-exclusivity assumption. The CI equation takes a different form depending on whether the two drugs are assumed to act through mutually exclusive or mutually non-exclusive mechanisms (the cross-term above is included only in the latter case) — picking the wrong assumption for the actual mechanism biases the result.
- Effect-level dependence. Because CI is typically calculated across multiple fa values, a combination can look synergistic at one effect level and additive or even antagonistic at another. Reporting a single CI value without specifying the effect level it corresponds to obscures this.
Software such as CompuSyn automates the median-effect and CI calculations and is widely used for exactly that reason, but automating the arithmetic does not remove the need for judgment about assay quality, dose-range selection, and whether the mutual-exclusivity assumption fits the drugs being tested. The practical takeaway: treat a reported CI alongside the raw dose-response curves and the isobologram it was derived from, not as a standalone verdict — a CI that looks synergistic on paper but rests on a poorly fit single-agent curve is not evidence of synergism, it’s evidence of a fitting problem.
Frequently Asked Questions
Is the Chou-Talalay method the only way to quantify drug combination synergy?
No. It is the most widely used method in pharmacology and cancer biology specifically because it ties synergy quantification directly to each drug’s own median-effect dose-response fit, but other frameworks exist (e.g., Bliss independence, Loewe additivity-based response-surface models) that make different assumptions about how independent drug effects should combine. Different frameworks can disagree at the margins on borderline cases, which is one more reason not to treat a single CI value as the final word.
Do I need a full dose-response curve for each drug, or just a couple of concentrations?
A full curve. The median-effect equation fit that CI depends on requires enough concentrations spanning the transition region of each drug’s own dose-response curve to estimate Dm and m reliably — a combination tested against only one or two single-agent doses cannot support a valid CI calculation, even if the arithmetic technically runs.
What does a fixed dose ratio mean in practice?
It means the two drugs are combined in a constant proportion (by concentration or dose) across the entire combination dilution series, rather than testing every possible pairing of doses. The ratio is usually chosen so each drug contributes a comparable share of the combined effect, most often based on the ratio of their individual IC50/ED50 values.
For the underlying single-agent dose-response concepts this method builds on, see CASRAI’s guides to the Hill coefficient in dose-response and binding curves, the Cheng-Prusoff equation for converting IC50 to Ki, and four-parameter logistic (4PL) curve fitting.
