DDI-Fusion: Human DDI Predictor

Predicts human DDI from in vitro CYP interaction data and chemical structure

Predicts human DDI (midazolam AUCR) using a proprietary algorithm that takes in vitro CYP interaction data (reversible inhibition, time dependent inhibition and induction) and chemical structure as inputs.
Customizable for alternative CYP3A4 substrates or alternative CYP enzymes.
chemPK™ v2 predicts key human pharmacokinetic parameters directly from structure. The predicted systemic and gut wall C_max values are used as inputs for DDI-Fusion.
DDI-Fusion has several advantages over the regulatory mechanistic static model in that:

it uses a data-driven optimization approach requiring fewer assumptions.
human C_max is predicted so no clinical in vivo data are required.

Provides early stage filter for directing chemistry and prioritizing screening.
Superior approach which uses PBPK model optimized from human clinical (in vivo) CYP3A4 data.

A tool for integrating in vitro ADME data with chemical structure to predict human DDI.

DDI-Fusion Input Requirements
Structural information input requirements Chemical structure., e.g., SMILES, mol or sdf Net charge at pH 7.4, calculable from pK_as Dosing information Dosing regimen of perpetrator (dose, route, frequency, duration) In vitro data Reversible (K_i or IC₅₀), TDI (K_I and k_inact) and induction (EC₅₀ and E_max) parameters Fraction unbound in plasma (f_u,p)

DDI-Fusion Data Delivery
Predicted AUCR of victim for oral delivery (plus liver and gut contributions)

How does it work?

DDI-Fusion integrates liver and gut PK data from chemPK™ with in vitro CYP interaction data in a KNIME workflow-based approach which executes the process illustrated in Figure 1.

Figure 1
DDI-Fusion workflow process.

KNIME can be downloaded easily and for free. Cyprotex can then provide the bespoke KNIME nodes (chemPK™ and DDI-Fusion) for the workflow.

Initially chemPK™ v2 predicts plasma and gut wall C_max values. These parameters are used along with the in vitro inhibition and /or induction data as inputs to the DDI-Fusion node. This node then predicts the AUCR_liver and AUCR_gut for estimation of the AUCR.

Calculation of AUCR by DDI-Fusion

DDI-Fusion predicts AUCR of a co-administered victim drug (e.g., midazolam) in the presence of the perpetrator drug (test article).

AUCR=	AUC of victim in presence of perpetrator
	AUC of victim in absence of perpetrator

If inhibitory effects dominate, AUCR>1
If induction effects dominate, AUCR<1

Additionally, DDI-Fusion predicts the contributions of DDI in (i) the gut (AUCR_gut), and (ii) the liver (AUCR_liver).

AUCR_total = AUCR_{gut wall} x AUCR_liver

Data

Data from DDI-Fusion

Figure 2
Comparison of (a). the mechanistic static model as recommended by regulatory authorities, and (b). Cyprotex’s DDI-Fusion model for predicting AUCR of reversible CYP3A4 inhibition against midazolam.

The data illustrate how the regulatory mechanistic static model over-predicts AUCR for inhibition leading to false positives and possible unwanted compound attrition.

	Regulatory Mechanistic Static Model	DDI-Fusion
R²	0.63	0.70
RMSE	3.67	2.53
GMFE	1.58	1.36

Table 1
Statistics comparing the regulatory mechanistic static model with Cyprotex's DDI-Fusion model.

We are looking for partners with whom we can further develop the DDI-Fusion model. If you would like to be involved then please get in touch.

Contact enquiries@cyprotex.com for a demonstration.

DDI-Fusion: Human DDI Predictor

Predicts human DDI from in vitro CYP interaction data and chemical structure

How does it work?

Data

Data from DDI-Fusion

Contact us to discuss your ADME Tox issues or request a quote

or fill out the form below: