DDPred allows the user to quantitatively predict the impact of pharmacokinetic drug-drug interactions that are mediated by cytochromes (CYPs), after oral administration.

The DDPred models currently consider the CYPs 3A4, 2D6, 2C9, 2C19 and 1A2.

The metrics are based on the Area Under the Concentration vs time curve (the AUC) of a substrate metabolized by one or several CYPs. Specifically, DDPred calculates the ratio of the AUC of the substrate that is coadministered with the interactor (inhibitor and/or inducer) to the AUC of the substrate administered alone. DDPred can identify multiple interactions involving several CYPs in a two-drug interaction. All types of inhibition (competitive, noncompetitive or mechanism-based) are considered. The AUC ratio is calculated with the interactor at steady-state, i.e. after repeated dosing.

The DDPred internal database currently includes about 150 substrates, 100 inhibitors and 20 inducers.

For users who are developing a new chemical entity (NCE) who wish to delineate the spectrum of major interactions with the NCE, DDPred calculates the AUC ratios for all combinations of the NCE with the compounds in the database.

Extended DDPred allows the user to calculate the impact of several situations on the exposure of drugs metabolized by CYPs:

1. CYP genetic polymorphism: the AUC ratio = the substrate AUC in a patient with the genetic variant / the substrate AUC in a homozygous wild type patient.
2. Drug interactions combined with genetic polymorphism: two kinds of AUC ratios are calculated, for the most frequent allelic combinations.
3. Cirrhosis: the AUC ratio = the substrate AUC in a cirrhotic patient / the substrate AUC in a non-cirrhotic patient. The AUC ratio is calculated for mild, moderate and severe cirrhosis according to the Child-Pugh class A, B or C.
4. Drug interactions in the presence of cirrhosis: two kinds of AUC ratios are calculated, for each class of cirrhosis.
5. Paediatric population: the clearance ratio = CL in child / CL in adult, where CL is expressed in L/h. The ratio is calculated according to body surface area and postnatal age.
6. Drug interactions in the paediatric population:  two kinds of AUC ratios are calculated according to postnatal age.

To use DDPred, at least one AUC ratio must have been measured in a clinical study of the interactions between the NCE and one compound in our the database.

If in vitro metabolism studies have shown that the NCE is a substrate of CYP3A4, there must be a phase 1 pharmacokinetic trial that has studied the interaction between the NCE and for example Ketoconazole, a strong inhibitor of CYP3A4. The geometric mean AUC ratio of the NCE is the only information needed by DDPred to calculate the AUC ratio with all other CYP3A4 inhibitors in the database.

If in vitro metabolism studies have shown that the NCE is an inhibitor of CYP2D6, there must be a phase 1 pharmacokinetic trial that has studied the interaction between the NCE and for example Dextromethorphan, a substrate that is almost completely metabolized by CYP2D6. The geometric mean AUC ratio of the NCE is the only information needed by DDPred to calculate the AUC ratio with all other CYP2D6 substrates in the database.

To predict the impact of genetic polymorphism, the user needs to know the major polymorphic CYP that’s involved in the metabolism of the NCE.

To predict the impact of cirrhosis, the user needs to know the unbound fraction in plasma and, preferably, the nature of the main binding protein.

DDPred does not currently take into account metabolism that involves other CYPs, such as 2C8 and 2B6.

DDPred calculations only apply for drugs that are administered orally. Note that changes after intravenous administration are less pronounced.

DDPred assumes the following:

1. The NCE kinetics are linear. The calculated AUC ratio is biased if there is a strong departure from linear kinetics.
2. The interactions are calculated only at the CYP level. If other mechanisms are present, such as changes in absorption due to gastric pH modification or competition with membrane transporters like OATP, PgP or others, the calculated AUC ratio will be biased.
3. Mutual interactions between two drugs are not taken into consideration. However, strong mutual interactions are rare.
4. Calculations are not precise when very strong inhibitors or substrates are involved.
5. When a substrate NCE is administered as a racemate, the enantiomers need to be considered separately. That is, the AUC of each enantiomer must be measured.
6. If the interaction involves a polymorphic CYP (2D6, 2C9, 2C19), the clinical interaction study must be carried out in a genetically homogenous population (e.g. CYP2D6*1*1 or CYP2D6*10*10). However, genetically homogenous subgroups with distinct genotypes may be considered.
7. To determine the impact of cirrhosis, the AUC ratios are biased downward if the liver extraction yield of the substrate is greater than 0.7.
8. No prediction can be made for preterm neonates using the paediatric module. Only data for term neonates can be used.

DDPred calculations are based on static (steady-state) equations developed for physiologically-based pharmacokinetic models. DDPred uses equations that are similar to those described in the FDA guidelines for the assessment of drug-drug interactions, although the equations are formulated differently. The main difference lies in the parameter values. Usually the parameters are determined by in vitro experiments that use, for example, microsomes or hepatocytes. With DDPred, all parameters have been determined in vivo in human subjects. This methodology eliminates problems due to discrepancies between in vitro and in vivo experiments that arise from non-Michaelian kinetics, adsorption, hetero-activation, product inhibition, other actions of metabolites on CYPs, time-dependant effects, and so on.

A whole body PBPK model allows one to simulate the drug concentration profile in all compartments (organs and tissues) and to take into account all mechanisms of interaction, provided that the corresponding parameters are known. The PBPK approach is suitable for interspecies scaling and therefore may be used in the preclinical phase of drug development. This approach is very powerful, but describing complex pharmacokinetic behavior requires a lot of work and validation. Even in simple cases, PBPK models frequently require the use of empirical “scaling factors” to match the predictions with the data. As a result, PBPK modeling is quite cumbersome and time consuming, and it requires a high level of expertise to produce meaningful information. The DDPred approach delivers a single parameter (an AUC ratio), but this information is often sufficient to make important decisions. Further, the AUC ratio is obtained very quickly and does not require a high level of expertise. However, it cannot be used in the preclinical stage. Hence, these two approaches are complementary rather than exclusive.

DDPred may also be used for an independent assessment of metabolic  DDIs, allowing to confirm PBPK results. This is especially valuable for DDIs involving multiple molecular species (e.g. metabolites or enantiomers of the interactor: bupropion, amiodarone, itraconazole, diltiazem, …).

The methodology behind all of the DDPred modules has been published in peer-reviewed journals, including Clinical Pharmacokinetics, AAPS Journal, Clinical Pharmacology and Therapeutics. In each paper, the DDPred approach was subjected to external validation that involved a comparison between DDPred predictions and published experimental data.

One published paper summarized the comparisons of drug-drug interactions as predicted by DDPred versus the interactions found in 628 interaction studies. It also compared the DDPred approach and the PBPK approach used by Simcyp® on 104 interaction studies. The predictive performance of DDPred was excellent and exceeeded that of the Simcyp® Simulator. Find the complete list of publications documenting DDPred in our news section.

DDPred requires clinical data as input. Hence, it cannot be used before phase 1. After the first clinical interaction study, it can be used to guide clinical development.

First, DDPred can be used to prioritize confirmatory interaction studies or genetic studies that need to be performed.

Second, DDPred can help anticipate the consequences of DDIs, genetic polymorphisms of CYPs, cirrhosis and immaturity. DDPred therefore facilitates the planning of phase 2 and phase 3 clinical studies by guiding the formulation of inclusion and exclusion criteria.

Third, DDPred can help determine appropriate prescription information.

Fourth, DDPred can help drug developers respond to health authorities, both during the approval process and after the process during phase 4 studies in terms of interpreting unexpected adverse events that may occur as the result of drug interactions, genetic polymorphisms, cirrhosis or immaturity.

Yes, this is possible. The data that are entered for each run are saved and exported in a file that can be reloaded later along with new data. In such cases, the predictions are updated after taking the new data into account.

No, it does not. All computations are performed on our secured servers in seconds.

The data you input on DDPred is not kept on our servers: once the computation is done, the user will automatically be prompted with a zip file to download which contains all the output, along with an encrypted file that you can use to make another run.

Yes, we can help users interpret their DDPred results. Registered users can benefit from such help by using the contact email. Complex issues may be discussed by phone.

DDPred courses may be organized either at the initiative of the staff of DDPred or to comply with the needs of registered users. Please use the contact email for further information.