ADMET Predictor® User Manual

CYP Induction

CYP Induction

Cytochrome P450 (CYP) induction refers to the process by which certain substances increase the activity of CYP enzymes, leading to an enhanced metabolic capacity. This phenomenon occurs when an inducer binds to nuclear receptors, such as the pregnane X receptor (PXR) or the aryl hydrocarbon receptor (AhR), which in turn upregulate the transcription of CYP genes. As a result, there is an increased synthesis of specific CYP enzymes, which accelerates the metabolism of drugs, xenobiotics, and endogenous substrates. This can significantly impact the pharmacokinetics of medications, potentially reducing their therapeutic efficacy and necessitating dosage adjustments. Induction can also lead to increased production of toxic metabolites, raising the risk of adverse effects. Common inducers include certain drugs, such as rifampicin and phenobarbital, as well as environmental chemicals like polycyclic aromatic hydrocarbons. Understanding CYP induction is crucial for predicting drug interactions and optimizing pharmacotherapy.

We have collected the literature data describing the CYP1A2 induction of 448 molecules. The data were initially retrieved from the DIDB database, and subsequently processed to assure appropriate data quality and consistency. The CYP induction measurements were carried out with various in-vitro techniques, generally based on the mRNA expression analysis, protein expression analysis or enzyme activity assays. The raw target values were expressed in terms of "induction fold" with respect to chosen baseline. However, due to the different experimental assays, these values were not always directly comparable. To overcome this problem we converted initial numerical "induction fold" quantity into a binary target, splitting the entire data set into classes of CYP1A2 inducers and not-inducers. All doubtful cases were carefully inspected with the help of source literature. The thorough analysis of these cases resulted in either unambiguous assignment of a compound to one of the classes, or removal from the consideration, if the literature did not resolve it unequivocally. This procedure led to a well-balanced dataset containing 246 CYP1A2 inducers and 202 non-inducers. With these data we developed an ANNE classifier which is characterized by the statistics presented below. The literature sources include 218 scientific publications and 92 NDA applications from FDA.

Model

Set

Negatives

Positives

Total

Correct

Concordance

Sensitivity

Specificity

CYP1A2_Ind

Training

170

210

380

297

78.2%

75.2%

81.8%

Test

32

36

68

54

79.4%

77.8%

81.2%