In drug development, particularly in preclinical research using cell lines, understanding drug response and accelerating drug combination discovery isn’t just about mixing two active ingredients — it’s about optimizing therapeutic strategies through effective drug combinations, uncovering the right drug interaction between them. A truly synergistic drug combination can amplify efficacy, reduce side effects, and overcome resistance mechanisms that often emerge during testing in cell lines and other preclinical models, or even later stages like clinical trials.
But how do we know when two drugs truly demonstrate drug synergy — and how do we quantify that using reliable synergy models that produce a meaningful synergy score?
That’s where drug synergy scoring models come in — serving as reference models to evaluate drug combinations more effectively in cancer cell lines and other biological models, and prioritize the most promising synergistic drug combinations for further testing. These models apply mathematical logic to generate a synergy score across cell lines and other in vitro systems, quantifying whether a combination demonstrates true drug synergy or just additive effects — and help researchers make smarter, faster decisions during screening campaigns.
In this article, we’ll break down the four most widely used synergy models — the Bliss model, Loewe, ZIP, and HSA — and show how they support more reliable, scalable drug combination screening.
Why We Need Scoring Models
Drug combination screening often reveals complex, nonlinear drug interactions. Some combinations improve outcomes, others show no added benefit, and a few can even interfere with each other.
Without a standardized scoring method, it’s difficult to interpret what those drug interaction results of each drug pair actually mean. Are two drugs genuinely synergistic when tested in cell lines? Or is the observed effect just additive?
Synergy models address this ambiguity by benchmarking observed combination responses and calculating a synergy score against a theoretical null model. This allows researchers to identify which drug pairs tested in cell lines merit further investment, progression to clinical trials, or mechanistic study.
Bliss Independence Model
The Bliss Independence model assumes that drugs act independently and calculates expected combined effects based on individual drug activity.
Key features:
- Best used when drugs have different mechanisms of action, where drug synergy can emerge from independent yet complementary effects
- Assumes probabilistic independence
- Tends to identify “excess over expected” effects, reflected in an elevated synergy score, as synergy
When to use it: Early-phase screening campaigns involving drugs that target distinct pathways.
Loewe Additivity Model
Loewe Additivity is based on the idea that combining similar drugs is like giving more of a single drug, assuming a predictable drug response to increasing concentrations. It compares the effect of a combination to the expected effect if the drugs were interchangeable.
Key features:
- Ideal for drug combinations with overlapping mechanisms, particularly in cancer cell lines where pathway redundancy is common
- Often used to determine dose equivalence
- Requires detailed dose-response matrix to compare drug effects across varying concentrations
When to use it: Later-stage analysis of drugs with similar modes of action or overlapping targets, where both dose-response curves and dose-response matrix enable accurate synergy drug combinations.
ZIP (Zero Interaction Potency) Model
The ZIP model integrates concepts from both Bliss and Loewe, offering nuanced synergy mapping across drug combinations tested in cell lines, using a response surface approach to model expected outcomes. It accounts for dose-response curves and drug potency across multiple concentrations.
Key features:
- Combines logic from both the Bliss model (independence) and the Loewe model (additivity logic)
- Supports 3D modeling of synergy landscapes derived from the underlying dose-response matrix
- Ideal for high-throughput combinatorial screening applications where a consistent synergy score is needed across hundreds of drug combinations and full dose-response matrix, enabling large-scale comparisons and deeper insights intro drug combination dynamics.
When to use it: Large-scale screening where detecting drug synergy across multiple concentrations is critical, scalable synergy scoring across many combinations and doses.
Highest Single Agent (HSA) Model
HSA is a simpler benchmark and reference model that compares the combination effect to the stronger of the two single agents.
Key features:
- Straightforward and easy to implement for generating a baseline synergy score comparison
- Doesn’t assume any drug interaction model, making it a neutral baseline for evaluation drug combinations
- Conservative estimate of synergy
When to use it: Quick filters or validation of a drug pair tested in cancer cell lines, especially when resources or time are limited.
AI-Augmented Scoring & Prediction
Modern screening platforms increasingly incorporate machine learning to enhance synergy prediction using historical data and established reference models based on patterns in the dose-response matrix, particularly in cancer cell lines, where complex drug interactions demand more nuanced analysis. AI and machine learning models can learn from past datasets on cell lines to predict which combinations are most likely to yield drug synergy — even before in vitro testing begins.
Applications include:
- Using machine learning to prioritize drug combinations based on predicted synergy score for wet-lab testing
- Modeling mechanism-specific interactions
- Integrating multi-omics or pathway-level data
When paired with traditional scoring methods, machine learning adds a predictive layer that boosts efficiency and reduces experimental costs and improves the prioritization of effective drug combinations.
Choosing the Right Model
Each model serves a unique purpose. Here’s a quick summary guide:
| Model | Best For | Assumptions |
| Bliss | Distinct mechanisms, early screens | Independent action |
| Loewe | Similar mechanisms, dose equivalence | Additivity |
| ZIP | High-throughput synergy score mapping | Hybrid of Bliss + Loewe |
| HSA | Conservative validation | No interaction assumption |
The best approach often involves using multiple reference models and comparing outputs across different drug combinations. Platforms like SynergyFinder make this easy by supporting all major scoring methods.
Conclusion
Drug synergy scoring models are essential tools for interpreting drug combinations and identifying drug interactions that matter. Whether you’re validating a hypothesis or screening at scale, choosing the right model helps generate a more accurate synergy score and improves decision-making, and reproducibility.
At Kyinno, our Drug Combination Screening Services integrate all four major models to deliver an accurate synergy score for each tested pair of drug combinations, along with AI-powered tools, to deliver actionable synergy insights into drug interactions and synergy scores you can trust.
→ Learn more about our screening platform or book a consultation today.