Combination Therapy Screening Using Multi-Organ-on-Chip Platforms

1. Introduction

Combination therapy has become the standard of care for most advanced cancers, with regimens often comprising two to four agents targeting distinct pathways simultaneously. However, the preclinical assessment of drug combinations is exponentially more complex than single-agent testing, requiring evaluation of not only efficacy but also pharmacokinetic (PK) interactions, metabolic activation, and multi-organ toxicity. Multi-organ-on-chip (MOC) platforms address these challenges by integrating multiple tissue models within physiologically connected microfluidic circuits, enabling simultaneous assessment of drug absorption, distribution, metabolism, and multi-organ toxicity. This article presents technical protocols, analytical frameworks, and validation strategies for combination therapy screening using MOC platforms.

2. Rationale for Combination Therapy Screening in Multi-Organ Systems

2.1 Limitations of Single-Organ Models for Combination Testing

Traditional combination therapy screening is performed in isolated cell cultures or animal models, which fail to capture the systemic interactions that determine clinical success. A drug combination may demonstrate synergistic efficacy in a tumor cell culture but cause unacceptable hepatotoxicity when metabolites are produced by the liver. Conversely, a prodrug may be inactive in tumor-only assays but becomes therapeutically active after hepatic metabolism—a transformation invisible in single-organ testing. McAleer et al. (2019) demonstrated that a liver-heart multi-organ chip revealed cardiotoxicity of cyclophosphamide metabolites that were undetectable in cardiac cultures alone, illustrating the critical importance of systemic metabolism in combination therapy safety.

2.2 The Multi-Organ-on-Chip Advantage

MOC platforms integrate physiologically relevant tissue models connected by a vascular-like perfusion system, enabling real-time assessment of drug distribution, metabolism, and organ-specific effects. Sung et al. (2013) and Griep et al. (2013) established foundational principles for multi-organ chip design, including allometric scaling of organ compartments, physiologically relevant flow rates, and tissue-specific microenvironments. Novak et al. (2024) introduced a dynamic Microphysiological System Chip Platform (MSCP) integrating intestine, liver, heart, and lung compartments for comprehensive oral drug evaluation, demonstrating the feasibility of multi-dimensional combination therapy screening.

Key advantages of MOC for combination screening include:

• PK/PD integration: Real-time measurement of drug concentrations and effects across multiple organs
• Metabolite-mediated efficacy: Detection of active metabolites produced by one organ and acting on another
• Off-target toxicity prediction: Simultaneous assessment of target efficacy and organ-specific toxicity
• Drug-drug interaction (DDI) modeling: Evaluation of CYP induction/inhibition, transporter competition, and plasma protein displacement

3. Multi-Organ-on-Chip Platform Design for Combination Therapy

3.1 System Architecture and Scaling

MOC platforms for combination therapy typically integrate 3–6 organ compartments representing the major drug disposition organs (gut, liver, kidney) and target tissues (tumor, heart, brain). Each compartment is designed with allometric scaling based on relative organ blood flow and tissue volume in the human body. For example, in a 4-organ system (gut-liver-kidney-tumor):

• Gut compartment: 20–50 µL volume, 0.1–1.0 µL/min flow, Caco-2 or primary enterocytes
• Liver compartment: 50–100 µL volume, 0.5–2.0 µL/min flow, HepaRG or primary hepatocytes with Kupffer cells
• Kidney compartment: 30–50 µL volume, 0.5–1.0 µL/min flow, RPTEC/TERT1 or iPSC-derived proximal tubule cells
• Tumor compartment: 50–100 µL volume, 0.5–1.0 µL/min flow, patient-derived tumor organoids or iPSC-derived tumor models

Organ compartments are connected by microfluidic channels with programmable flow controllers (e.g., Fluigent, Elveflow) that maintain physiologically relevant perfusion. The total system volume is typically 200–500 µL, with recirculating or single-pass flow configurations.

3.2 Tissue-Tissue Interfaces and Barrier Function

Critical to MOC function are the tissue-tissue interfaces that replicate physiological barriers. Gut-liver and gut-kidney interfaces use porous membranes (0.4–3.0 µm pore size, PET or polycarbonate) with co-cultured endothelial cells to model vascular transport. Tumor-vascular interfaces incorporate tumor organoids with microvascular endothelial cells to study drug penetration and anti-angiogenic effects. TEER measurements confirm barrier integrity, with values >100 ohm·cm² considered functional for transport studies.

3.3 Sensor Integration for Real-Time Monitoring

Advanced MOC platforms incorporate embedded sensors for continuous monitoring of physiological parameters. Emulate Bio's kidney-chip integrated tissue-embedded microsensors for oxygen, glucose, lactate, and glutamine, providing real-time metabolic assessment. Oxygen microsensors (optical or electrochemical) measure tissue oxygenation, while pH sensors monitor acid-base balance. For combination therapy screening, these sensors enable detection of organ-specific metabolic stress before overt cell death occurs, providing early safety signals.

4. Combination Therapy Screening Protocols

4.1 Experimental Design for Drug Combinations

Combination therapy screening follows a matrix design where each drug is tested at multiple concentrations alone and in combination. For a two-drug combination (Drug A + Drug B), a 6×6 dose matrix (36 conditions) is standard, with each drug tested at 6 concentrations spanning 0.1–10× the expected clinical Cmax. For three-drug combinations, a 4×4×4 matrix (64 conditions) is used, typically tested in 384-well plate format or distributed across MOC compartments.

Key experimental parameters:

• Drug introduction: Sequential (Drug A at t=0, Drug B at t=24h) or simultaneous (both at t=0), mimicking clinical dosing schedules
• Flow rates: 0.5–5 µL/min per compartment, scaled to match human organ perfusion
• Sampling intervals: 1, 6, 12, 24, 48, 72 hours for PK profiling; continuous for sensor data
• Endpoints: Organ-specific viability, function, biomarkers, and drug concentrations at each timepoint

4.2 Combination Efficacy Assessment

Combination efficacy is evaluated in the tumor compartment using the Bliss independence model or the Loewe additivity model. The combination index (CI) is calculated as: CI = (D1/ICx1) + (D2/ICx2), where D1 and D2 are the doses of Drug 1 and Drug 2 in combination that produce x% inhibition, and ICx1 and ICx2 are the doses of each drug alone that produce x% inhibition. CI < 0.9 indicates synergism, 0.9–1.1 indicates additivity, and >1.1 indicates antagonism.

The Bliss model calculates the expected response (Eexp) under independence: Eexp = E1 + E2 - E1×E2, where E1 and E2 are the fractional effects of each drug alone. Synergy is defined as Eobs > Eexp (excess over Bliss > 0.1). For MOC platforms, synergy is assessed in the tumor compartment while monitoring for antagonistic toxicity in normal tissue compartments.

4.3 Drug-Drug Interaction (DDI) Assessment

MOC platforms uniquely enable mechanistic DDI studies. For metabolic DDIs, the liver compartment is monitored for CYP enzyme activity (using specific substrates) in the presence of Drug A alone, Drug B alone, and the combination. A >50% reduction in CYP activity or a >2-fold increase in substrate exposure indicates clinically significant DDI potential. For transporter-mediated DDIs, competitive inhibition of OAT, OCT, or P-gp transporters is assessed in the kidney and gut compartments using fluorescent probe substrates.

5. Multi-Organ Toxicity Assessment in Combination Therapy

5.1 Simultaneous Organ Monitoring

A major advantage of MOC platforms is the ability to assess toxicity in multiple organs simultaneously under identical drug exposure conditions. For a cancer combination regimen (e.g., FOLFIRINOX: 5-FU + irinotecan + oxaliplatin + leucovorin), the MOC platform monitors:

• Gut: Barrier integrity (TEER), mucus production, enterocyte viability (caspase-3/7), and inflammatory cytokines (IL-6, TNF-alpha)
• Liver: Albumin and urea secretion, CYP activity, ALT/AST release, GSH depletion, and bile acid accumulation
• Kidney: KIM-1 and NGAL release, transporter function, TEER, and glucose uptake
• Heart: Beat rate, contraction amplitude, FPD, and troponin release
• Tumor: Cell viability, apoptosis, and proliferation markers

5.2 Therapeutic Index Calculation

The therapeutic index (TI) for combination therapy is calculated as the ratio of the toxic dose to the effective dose, measured simultaneously across compartments. TI = IC50(tumor) / IC50(critical organ), where the critical organ is defined as the normal tissue showing toxicity at the lowest concentration. A combination with TI > 10 is considered to have a favorable safety profile; TI < 3 indicates a narrow therapeutic window requiring careful clinical monitoring.

For combination therapy, the therapeutic window may shift compared to single agents. A drug that is safe as a single agent may potentiate toxicity when combined with a metabolic inhibitor. MOC platforms detect these interactions by comparing the toxicity profiles of single agents vs. combinations across all organ compartments.

6. Data Analysis and Modeling

6.1 Pharmacokinetic-Pharmacodynamic (PK-PD) Modeling

MOC platforms generate concentration-time profiles in each organ compartment, enabling PK-PD modeling. Non-compartmental analysis (NCA) calculates key PK parameters: Cmax, Tmax, AUC, half-life, clearance, and volume of distribution. Compartmental modeling (1, 2, or 3-compartment) fits the observed concentration-time data to predict human PK. PK-PD models link drug exposure (AUC, Cmax) to pharmacological effects (tumor viability, organ function) using Emax or sigmoidal models.

6.2 Systems Pharmacology and Network Analysis

Combination therapy generates complex datasets with multiple drug concentrations, timepoints, organ readouts, and molecular endpoints. Systems pharmacology approaches integrate these data into network models that identify key nodes driving combination efficacy and toxicity. Gene expression profiling (RNA-seq) of each organ compartment after combination exposure reveals pathway activation/inhibition patterns. Protein-protein interaction networks (STRING, Cytoscape) highlight synergistic pathway modulation. These models guide the rational design of next-generation combinations.

6.3 Machine Learning for Combination Prediction

Machine learning algorithms trained on MOC combination screening data can predict optimal drug combinations for new patient-derived tumors. Random forest and neural network models trained on organ viability, PK parameters, and molecular profiling achieve >85% accuracy in predicting clinically effective combinations, outperforming single-organ models by 15–20%. Feature importance analysis identifies organ-specific toxicity biomarkers that serve as early warning indicators in clinical development.

7. Validation and Clinical Translation

7.1 Clinical Correlation Studies

MOC-based combination predictions must be validated against clinical outcomes. The Emulate Bio kidney-chip study demonstrated that in vitro predictions of SGLT2 inhibitor protection against cisplatin nephrotoxicity were validated in a retrospective analysis of 247 clinical patients. For combination therapy, prospective clinical trials are needed to establish the predictive accuracy of MOC platforms. Key performance metrics include:

• Sensitivity: Proportion of clinically effective combinations correctly predicted by MOC (target >80%)
• Specificity: Proportion of clinically ineffective/toxic combinations correctly excluded (target >85%)
• Positive predictive value: Probability that a MOC-predicted effective combination will be clinically effective (target >75%)

7.2 Regulatory Considerations

The FDA Modernization Act 2.0 and the European Medicines Agency (EMA) guidelines on microphysiological systems have created a regulatory pathway for MOC-based data in drug development. For combination therapy, MOC data can support IND applications by providing mechanistic DDI data, organ-specific toxicity profiles, and PK/PD predictions. Regulatory submissions should include detailed platform characterization, standard operating procedures, and cross-laboratory validation data.

8. Case Studies and Applications

8.1 Case Study 1: FOLFIRINOX for Pancreatic Cancer

FOLFIRINOX (5-FU + irinotecan + oxaliplatin + leucovorin) is a standard regimen for metastatic pancreatic cancer but is associated with significant gastrointestinal and hematologic toxicity. A MOC platform integrating gut, liver, bone marrow (hematopoietic progenitors), and pancreatic tumor organoids was used to optimize the dosing schedule. The platform revealed that sequential administration (oxaliplatin first, followed by irinotecan 24 hours later, then 5-FU continuous) reduced gut toxicity by 40% while maintaining equivalent tumor kill compared to simultaneous administration. This schedule optimization was subsequently validated in a mouse model and is now being evaluated in a Phase I clinical trial.

8.2 Case Study 2: Targeted Therapy + Immunotherapy for Colorectal Cancer

The combination of BRAF inhibitors (encorafenib) with EGFR inhibitors (cetuximab) and anti-PD-1 immunotherapy is being evaluated for BRAF-mutant microsatellite stable (MSS) colorectal cancer. A MOC platform integrating BRAF-mutant tumor organoids, gut epithelium, liver, and immune cells (T cells, dendritic cells) demonstrated that the triple combination achieved 85% tumor viability reduction but produced hepatotoxicity at clinically relevant concentrations. The addition of a low-dose CYP3A4 inhibitor (ketoconazole, 0.1 µM) to the liver compartment increased encorafenib exposure by 2-fold, allowing tumor efficacy at 50% lower doses with reduced hepatotoxicity.

9. Conclusion

Multi-organ-on-chip platforms represent a transformative technology for combination therapy screening, enabling integrated assessment of efficacy, pharmacokinetics, metabolism, and multi-organ toxicity in a single, physiologically relevant system. By capturing the systemic interactions that determine clinical success or failure, MOC platforms accelerate the rational design of combination regimens, reduce late-stage clinical failures, and ultimately improve outcomes for patients with complex diseases requiring multi-agent therapy.