Regulatory

FDA DDI Risk Assessment: The R1/R2/R3 Framework Explained

Pharazi.ai TeamMarch 5, 20268 min read

Why DDI Risk Assessment Dictates Clinical Pipelines

Severe drug-drug interactions (DDIs) are responsible for lethal adverse events, black-box warnings, and extreme regulatory scrutiny. The FDA demands quantitative DDI evaluation as early as the IND stage.

The 2020 FDA DDI Guidance solidifies three primary static prediction models:

The R1 Model (Reversible Inhibition)

R_{1} = 1 + \frac{[ I ]}{K _{i}}

Where:

$[I]$ = Maximal inhibitor concentration at the enzyme site (usually $C_{ma x, u}$ )
$K_{i}$ = Reversible inhibition constant from in vitro kinetic assays

DDI flag triggered. FDA recommends further modeling or in vivo clinical validation.

Low risk. A dedicated clinical DDI study is unlikely to be mandated.

The R2 Model (Time-Dependent Inhibition)

When irreversible, mechanism-based inhibition is observed:

R_{2} = \frac{k _{o b s} + k _{d e g}}{k _{d e g}}

k_{o b s} = \frac{k _{ina c t} \cdot [ I ]}{K _{I} + [ I ]}

Parameter	Definition
$k_{ina c t}$	Maximum enzyme inactivation rate
$K_{I}$	Concentration for half-maximal inactivation
$k_{d e g}$	Endogenous CYP enzyme degradation rate

[!CAUTION]
$R_{2} \geq 1.25$ signals strong TDI risk — plan a clinical DDI trial immediately. R2 is critical for detecting mechanism-based inhibitors like erythromycin (CYP3A4) where enzymatic destruction persists for **days** after the inhibitor clears.

The AUCR Net Effect Model

When both reversible and TDI effects coexist:

A U C R = \frac{1}{[ A _{g} \cdot B _{g} \cdot ( 1 - F _{g} ) + F _{g} ] \cdot [ \frac{f _{m}}{A _{h} \cdot B _{h}} + ( 1 - f _{m} )]}

The Tiered Decision Pipeline

Screen with R1

$R_{1} < 1.02$ → Safe, halt investigation
$R_{1} \geq 1.02$ → Proceed to deeper modeling

Scan for TDI Markers

$R_{2} < 1.25$ → TDI dismissed
$R_{2} \geq 1.25$ → Fuse TDI with reversible models

Execute AUCR Prediction

AUCR Range	Interpretation
< 1.25	Safe — clinical waiver likely
1.25–2	Weak inhibitor — study suggested
2–5	Moderate inhibitor — trial required
≥ 5	Strong inhibitor — black-box labeling probable

Common Pitfalls

[!WARNING]

Using total $C_{ma x}$ instead of mandated unbound $C_{ma x, u}$ causes massive overestimations

Forgetting active metabolites that inhibit independently of parent

Static models are intentionally hyper-conservative — PBPK modeling may show lower true risk

Try It on Pharazi.ai

Run your data through the DDI Risk Assessment Calculator or the DDI Magnitude Calculator.

References

FDA. In Vitro Drug Interaction Studies. January 2020.
Vieira MLT, et al. Clin Pharmacol Ther. 2014;95(2):189–198.

DDIdrug interactionsFDA guidanceCYP inhibitionR1 R2 R3regulatory complianceAUCR

All articles