When treating complex diseases like cancer or multi-drug resistant infections, combining multiple therapeutic agents often yields better results than increasing the dose of a single drug. However, simply observing that a combination is effective does not mean the drugs are truly working together. If Drug A kills 30% of a pathogen population and Drug B kills 30%, does a combined effect of 50% mean they are cooperating, or are their effects merely overlapping?
To design optimized therapies, researchers must distinguish between simple additivity (where drugs act independently or substitute for one another) and true synergy (where the combined effect is greater than the sum of its parts). Quantifying this relationship requires rigorous mathematical frameworks that map experimental data to theoretical baselines of independence.
Mapping the Interaction Space via Checkerboard Assays
Before applying mathematical models, researchers must systematically collect data across a wide range of doses. The standard experimental design for this is the checkerboard assay (or concentration matrix screening).
In a checkerboard assay, two drugs are serially diluted across a two-dimensional grid (typically a 96-well plate). One drug increases in concentration along the x-axis, while the other increases along the y-axis. This creates a comprehensive map of every possible dose combination, allowing researchers to observe how the presence of one compound alters the potency of another.
As noted by Radic-Sarikas et al., screening pairs across a full concentration matrix is essential for capturing non-linear drug behaviors. This technique is widely used across antimicrobial research. For instance, Schwarz et al. and Bange in In Vitro Synergy of Isavuconazole Combined With Colistin Against Common Candida Species utilized broth microdilution checkerboards to evaluate antifungal combinations against Candida species, providing the raw data needed to calculate precise interaction metrics.
The Fractional Inhibitory Concentration Index
Once checkerboard data is collected, the most common method for evaluating antimicrobial combinations is Fractional Inhibitory Concentration Indexing (FICI).
The intuition behind FICI is straightforward: it measures how much of a drug's required dose can be reduced when it is paired with another agent. To calculate FICI, the Minimum Inhibitory Concentration (MIC) of each drug in combination is divided by its MIC when acting alone:
$$\text{FICI} = \frac{\text{MIC}_{\text{A in combination}}}{\text{MIC}_{\text{A alone}}} + \frac{\text{MIC}_{\text{B in combination}}}{\text{MIC}_{\text{B alone}}}$$
If the two drugs substitute for one another perfectly on a 1:1 basis, the sum of these fractions will equal $1$ (additivity). If the drugs actively hinder each other, the fraction will exceed $1$ (antagonism). However, if the drugs are synergistic, they amplify each other's potency, meaning far lower concentrations are required to achieve the same inhibitory effect. Typically, an $\text{FICI} \le 0.5$ is the strict threshold used to declare true synergy.
This metric allows researchers to mathematically validate when an adjuvant restores susceptibility to a failing drug. For example, Eldesouky et al., She et al., Xing et al., Yuan et al., Chu et al., and Salama et al. all rely on FICI calculations to prove that combining repurposed non-antimicrobial compounds or natural extracts with conventional antibiotics or antifungals can successfully bypass microbial resistance mechanisms.
The Median-Effect Principle and Combination Index
While FICI works well for binary outcomes like microbial growth inhibition (growth vs. no growth), oncology and complex signaling pathways require a model that accounts for continuous, sigmoidal dose-response curves. For these applications, researchers turn to the Chou-Talalay method, which is built upon the median-effect principle.
The median-effect principle linearizes complex dose-response curves by relating the dose of a drug to its fractional effect (e.g., percent of tumor cells killed). By mapping the individual curves of Drug A and Drug B, the method derives a theoretical baseline for how the two curves should merge if their interaction is purely additive. This baseline is used to calculate a Combination Index (CI).
As explained by Abécassis et al. and Xu et al., the CI quantifies the discrepancy between the observed drug response and the expected additive effect across multiple concentration ranges:
- $\text{CI} < 1$ indicates synergy (the combination is more effective than expected).
- $\text{CI} = 1$ indicates additivity.
- $\text{CI} > 1$ indicates antagonism.
Combination Index (CI) Spectrum:
[ Synergy (CI < 1) ] <---- [ Additivity (CI = 1) ] ----> [ Antagonism (CI > 1) ]
This mathematical framework is a cornerstone of oncology research. Deng et al. explicitly used the Chou-Talalay method to demonstrate the synergy of pyrotinib and adriamycin in breast cancer. Similarly, the CI metric allowed Kong et al., Xiao et al., Sturm et al., Rong et al., and Du et al. to objectively prove that targeted inhibitors can sensitize cancer cells to chemotherapy, justifying lower clinical dosages and reducing systemic toxicity.
To visualize these interactions across an entire matrix, researchers often construct an isobologram. As demonstrated by Pei in Sorafenib Inhibits Doxorubicin-Induced PD-L1 Upregulation to Improve Immunosuppressive Microenvironment in Osteosarcoma, an isobologram plots lines of equal therapeutic effect. If the combination curve bows inward toward the origin, it visually confirms that a lower total dose is achieving the target effect, providing an intuitive graphical counterpart to the calculated CI.
Alternative Baselines and High-Dimensional Scaling
While FICI and CI dominate the literature, other mathematical models offer unique advantages depending on the experimental context:
- Observed vs. Expected Ratios: In toxicology and environmental science, researchers like Nelson et al. and Song et al. compare observed mixture toxicities directly against calculated expected values. Guo et al. formalized this via the Model Deviation Ratio (MDR), which quantifies how far a multi-chemical "cocktail" deviates from standard additivity models as concentrations fluctuate.
- Calculated Coefficients (Q-values): In multi-agent herbal or traditional medicine formulations, Chen et al. utilize threshold-based Q-values to distinguish true adjuvant sensitization from simple cumulative toxicity.
- Partial Derivatives: To capture how synergy shifts dynamically across different dose ranges within a biological pathway, Han et al. introduced a second-order partial derivative of the system's survival ratio, tracking the sensitivity of drug interactions as a smooth mathematical surface rather than a single static index.
Overcoming the Curse of Dimensionality
As medicine moves toward three- or four-drug combinations, standard checkerboard assays face an exponential explosion of required experiments—a 10-dose grid for two drugs requires 100 wells, but four drugs would require 10,000 wells.
To solve this, Brennan et al. developed a normalized diagonal sampling design paired with a minimax synergy metric. Instead of sampling the entire high-dimensional space, this design strategically samples along normalized diagonals where individual drug concentrations match specific clinical benchmarks. The minimax metric then identifies the smallest maximum concentration required among effective combinations, allowing high-dimensional synergy to be mapped with orders of magnitude fewer experiments.
To streamline these complex, multi-model calculations, modern researchers frequently leverage automated software platforms like SynergyFinder. Used by Song et al., these tools integrate multiple mathematical baselines simultaneously, ensuring that whether a study is evaluating a natural extract, an antibiotic, or a triple-therapy cancer regimen, the quantification of synergy remains precise, reproducible, and biologically meaningful.
Key Foundational Work
A formal model for analyzing drug combination effects and its application in TNF-α-induced NFκB pathway (2010) — contributed 2 concept elements to this direction.
Histone deacetylase inhibitors sensitize tumour cells for cytotoxic effects of natural killer cells (2008) — contributed 1 concept elements to this direction.