Computational design and evaluation of novel thienopyrimidine PI3Kδ inhibitors for diffuse large b-cell lymphoma: QSAR, docking, molecular dynamics, and density functional theory studies

Abstract

Phosphoinositide 3-kinase delta (PI3Kδ) is a key regulator of B-cell receptor signalling and a validated therapeutic target in diffuse large B-cell lymphoma (DLBCL). The development of selective and pharmacokinetically balanced PI3Kδ inhibitors remains a clinical priority. An integrated computational workflow was applied to a curated dataset of 29 thienopyrimidine derivatives with reported pIC₅₀ values. A quantitative structure–activity relationship (QSAR) model was developed using multiple linear regression with genetic algorithm-based descriptor selection. Lead prioritization combined QSAR prediction, molecular docking, ADMET profiling, molecular dynamics (MD) simulations, MM-GBSA binding free-energy estimation, and density functional theory (DFT) calculations. The QSAR model demonstrated strong internal and external validation (R 2 = 0.868; Q 2 _LOO = 0.810; R 2 _test = 0.709) and identified key electronic, topological, and lipophilic determinants of PI3Kδ inhibition. Compound 14 was selected as a balanced scaffold and structurally optimized to generate four derivatives. Among them, 14D3 exhibited the highest predicted activity (pIC 50 = 8.387), favourable docking affinity (-9.6 kcal/mol), high predicted intestinal absorption, and non-AMES toxicity. MD simulations confirmed stable binding within the PI3Kδ active site, supported by persistent hydrogen bonding and hydrophobic interactions. MM-GBSA and DFT analyses indicated favourable binding energetics and enhanced electronic stability relative to the reference inhibitor copanlisib. Compound 14D3 emerges as a promising PI3Kδ inhibitor candidate, demonstrating the value of an integrated computational strategy for rational design and prioritization of targeted therapeutics for DLBCL. • A validated QSAR model was developed to predict the activity of thienopyrimidine-based PI3Kδ inhibitors. • Structure-based docking and molecular dynamics simulations elucidated key binding interactions and stability within the PI3Kδ active site. • Rational scaffold optimization led to the design of a promising derivative with improved predicted potency and drug-like properties. • Density functional theory analyses provided insight into the electronic stability and reactivity of the optimized compound. • The integrated in silico workflow offers a robust framework to support future experimental development of PI3Kδ inhibitors for DLBCL.