Using the Variable-Nearest Neighbor Method To Identify P-Glycoprotein Substrates and Inhibitors

Area of Science:

• Pharmacology
• Computational Chemistry
• Drug Discovery

Background:

• Permeability glycoprotein (Pgp) is a key transporter influencing drug pharmacokinetics.
• Overexpression of Pgp contributes to multidrug resistance in cancer by reducing intracellular drug concentrations.
• Identifying Pgp substrates and inhibitors is vital for effective cancer chemotherapy.

Purpose of the Study:

• To develop quantitative structure-activity relationship (QSAR) models for predicting Pgp substrates and inhibitors.
• To utilize a variable-nearest neighbor (v-NN) method for robust molecular similarity calculations.
• To establish models with well-defined applicability domains for reliable predictions.

Main Methods:

• Development of QSAR models using publically available data.
• Application of the variable-nearest neighbor (v-NN) method based on molecular structural similarity.
• Validation of models using external datasets of candidate Pgp substrates and inhibitors.

Main Results:

• The v-NN models achieved high accuracy (>80%) and reliability (κ values >0.60) in predicting Pgp substrates and inhibitors.
• Performance of v-NN models was comparable or superior to other modeling approaches.
• The developed models demonstrated a well-defined applicability domain for accurate predictions.

Conclusions:

• The v-NN QSAR models provide accurate and reliable predictions of Pgp substrates and inhibitors.
• The v-NN method is computationally efficient and adaptable to new data, facilitating model updates.
• These models hold significant potential for advancing cancer drug discovery and overcoming multidrug resistance.