Scholarlink

<p>Newly approved kinase inhibitors (2024–2025) have limited post-approval safety data. We developed a weighted ToxPi framework to prioritize multi-endpoint liabilities spanning hERG cardiotoxicity, DILI, Ames genotoxicity, and CYP3A4-mediated drug-drug interaction potential. Thirteen compounds (12 KIs approved in 2024–2025 plus imatinib as a historical reference anchor) were profiled using ADMETlab 3.0, vNN-ADMET, and SwissADME (formula inputs) and integrated into a weighted composite (wToxPi = 0.30 × hERG + 0.30 × DILI + 0.20 × Ames + 0.20 × CYP3A4). Robustness was assessed across weighting schemes and sensitivity analyses addressing fusion-recoding scenarios, CYP3A4 platform asymmetry, and combination-regimen confounding. DILI signal saturation dominated the cohort: 12 of 13 compounds showed DILIcon ≥ 0.80, limiting hepatic discrimination and shifting differentiation to hERG and Ames variability. Sevabertinib (0.860), Zongertinib (0.840), and Lazertinib (0.775) ranked highest, whereas Mirdametinib (0.080) was the structural outlier. Ranking remained stable: ADMETlab-only and thresholded majority-vote alternatives preserved the principal structure (Spearman <i>ρ</i> = 0.780 and 0.812), whereas soft-label pseudo-probability mapping showed near-identity concordance (<i>ρ</i> = 0.978); a broader sensitivity grid (<i>p</i> = 0.60–0.90) confirmed structural stability (<i>ρ</i> = 0.929–0.995). Exploratory FAERS findings provided supportive post-marketing context. This framework provides transparent, reproducible, hypothesis-generating prioritization of newly approved KIs. Class-level DILI saturation reinforces the need for close hepatic surveillance in pharmacovigilance interpretation regardless of composite rank; hERG and Ames provide the main inter-compound discriminatory signal. Sevabertinib, Zongertinib, and Lazertinib ranked highest, whereas Mirdametinib was the structural outlier; for Inavolisib and Tovorafenib, endpoint-specific hepatic surveillance remains warranted irrespective of composite rank. The model should support pharmacovigilance prioritization rather than replace clinical monitoring or individualized risk assessment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Class-level DILI saturation and hERG-driven risk stratification among 2024–2025 FDA-approved kinase inhibitors: an in silico multi-platform safety analysis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月2日

查看全文
  • Nihan Küçük Yılmaz,
  • Veysel Baskın

摘要

Newly approved kinase inhibitors (2024–2025) have limited post-approval safety data. We developed a weighted ToxPi framework to prioritize multi-endpoint liabilities spanning hERG cardiotoxicity, DILI, Ames genotoxicity, and CYP3A4-mediated drug-drug interaction potential. Thirteen compounds (12 KIs approved in 2024–2025 plus imatinib as a historical reference anchor) were profiled using ADMETlab 3.0, vNN-ADMET, and SwissADME (formula inputs) and integrated into a weighted composite (wToxPi = 0.30 × hERG + 0.30 × DILI + 0.20 × Ames + 0.20 × CYP3A4). Robustness was assessed across weighting schemes and sensitivity analyses addressing fusion-recoding scenarios, CYP3A4 platform asymmetry, and combination-regimen confounding. DILI signal saturation dominated the cohort: 12 of 13 compounds showed DILIcon ≥ 0.80, limiting hepatic discrimination and shifting differentiation to hERG and Ames variability. Sevabertinib (0.860), Zongertinib (0.840), and Lazertinib (0.775) ranked highest, whereas Mirdametinib (0.080) was the structural outlier. Ranking remained stable: ADMETlab-only and thresholded majority-vote alternatives preserved the principal structure (Spearman ρ = 0.780 and 0.812), whereas soft-label pseudo-probability mapping showed near-identity concordance (ρ = 0.978); a broader sensitivity grid (p = 0.60–0.90) confirmed structural stability (ρ = 0.929–0.995). Exploratory FAERS findings provided supportive post-marketing context. This framework provides transparent, reproducible, hypothesis-generating prioritization of newly approved KIs. Class-level DILI saturation reinforces the need for close hepatic surveillance in pharmacovigilance interpretation regardless of composite rank; hERG and Ames provide the main inter-compound discriminatory signal. Sevabertinib, Zongertinib, and Lazertinib ranked highest, whereas Mirdametinib was the structural outlier; for Inavolisib and Tovorafenib, endpoint-specific hepatic surveillance remains warranted irrespective of composite rank. The model should support pharmacovigilance prioritization rather than replace clinical monitoring or individualized risk assessment.