A Phase 2A Randomized Double-Blind Placebo-Controlled Trial to Evaluate the Efficacy and Safety of Varoglutamstat (PQ912) in Patients With Early Alzheimer's Disease With a Stage-Gate to Phase 2B (VIVA-MIND)

This is a Phase 2A multi-center, randomized, double blind, placebo-controlled, parallel group study of varoglutamstat, with a stage gate to Phase 2B.
In Phase 2A there will be adaptive dosing evaluation of three dose levels with exposure to varoglutamstat or placebo for a minimum of 24 weeks, with preliminary evaluation of both cognitive function and pharmacodynamic changes on EEG spectral analysis in approximately 180 participants.
In the event that the stage gate for Phase 2B is reached, then Phase 2B will assesses efficacy and longer-term safety in a larger study group, i.e., 414.

开始日期2021-11-15

申办/合作机构

Vivoryon Therapeutics NV

[+1]

NCT02389413

/ Completed临床2期

A Phase 2A Multicentre, Randomised, Double Blind, Placebo-Controlled, Parallel-Group Safety and Tolerability Study of PQ912 in Subjects With Early Alzheimer's Disease

The aim of this study is to evaluate the safety, tolerability and preliminary efficacy of PQ912 in subjects with Mild Cognitive Impairment (MCI) due to Alzheimers Disease (AD) or mild dementia due to AD.

开始日期2015-03-01

申办/合作机构

Vivoryon Therapeutics NV

[+1]

NCT02190708

/ Completed临床1期

A Phase I, Open Label Study to Assess the Effects of PQ912 on the Pharmacokinetics of Midazolam and Omeprazole in Healthy Male Subjects

Midazolam is a rapid-acting benzodiazepine, with a short half-life (approximately 1.9 hours) and is primarily metabolised by CYP3A.
Omeprazole is a selective proton pump inhibitor substrate used to reduce gastric acid secretion. Omeprazole is primarily metabolised by CYP2C19.
Midazolam and omeprazole are both used as probe drugs in clinical pharmacology studies to evaluate clinical CYP3A and CYP2C19 drug interactions, respectively. Furthermore the EMA and the FDA guidance on drug interactions recommend the use of these drugs for such evaluations.
The aim of this study is to assess the effect of PQ912 on the PK of midazolam and omeprazole. In vitro studies have demonstrated that PQ912 inhibits several CYP enzymes, including CYP3A4 and CYP2C19 and at the expected exposure levels in patients, has the potential to inhibit these enzymes in-vivo. This study is therefore planned to investigate the potential changes in the PK of midazolam and omeprazole due to the effect of PQ912 at steady-state. In clinical practice it is likely that co-administration of PQ912 with other drugs that are metabolised via the CYP3A and/or CYP2C19 enzymes will occur. This study will provide important information for the requirement of dose adjustments or contraindications in these circumstances.

开始日期2014-06-01

申办/合作机构

Vivoryon Therapeutics NV

[+1]

100 项与 Varoglutamstat 相关的临床结果

登录后查看更多信息

100 项与 Varoglutamstat 相关的转化医学

登录后查看更多信息

100 项与 Varoglutamstat 相关的专利（医药）

登录后查看更多信息

项与 Varoglutamstat 相关的文献（医药）

2026-05-01·JPAD-Journal of Prevention of Alzheimers Disease

Validation of a novel cognitive-functional outcome measure optimized for early Alzheimer’s Disease: Evidence from the VIVA-MIND trial

Article

作者: Wassmann, Tanja ; Meyer, Antje ; Schaeffer, Michael ; MacKelfresh, Andrew J ; Duehring, Jasmin A ; Wenzkowski, Christine ; Revta, Carolyn ; Jacobs, Diane M ; Schell-Mader, Sylvia ; Edland, Steven D ; Feldman, Howard H ; Salmon, David P

BACKGROUND:

Cognitive-functional composite measures are increasingly used as primary efficacy endpoints in early Alzheimer's disease (AD) trials, where greater sensitivity to decline can improve trial efficiency and reduce sample size requirements.

OBJECTIVES:

To compare sensitivity to decline of the Cognitive Functional Component 2 (CFC2), a novel cognitive-functional composite measure described by Raghavan et al. (2013), against the Clinical Dementia Rating - Sum of Boxes (CDR-SB) and other standard cognitive and functional outcomes including MMSE, FAQ, ADAS Cog 13 and ADCOMS using prospective randomized clinical trial data.

DESIGN:

The VIVA-MIND trial was a phase 2A/2B randomized controlled trial investigating the safety and efficacy of varoglutamstat in patients with mild cognitive impairment and mild dementia due to AD.

SETTING:

The VIVA-MIND trial was conducted between 2021-2024. It was prematurely terminated in mid-2024 by the study sponsor.

PARTICIPANTS:

This secondary analysis uses data from 98 participants in the modified intention-to-treat population from the VIVA-MIND trial with complete neuropsychological test data.

MEASUREMENTS:

Standard power calculations informed by parameters estimated from linear mixed-effects models were used to determine the relative efficiency of outcome measures.

RESULTS:

The CFC2 was more sensitive to decline than the CDR-SB in this population. Use of the CFC2 would yield a 15% reduction in required sample size relative to the CDR-SB. Application of an optimal weighting scheme further improved the sensitivity of the CFC2.

CONCLUSIONS:

Practically significant differences in the efficiency of clinical trials in early AD may be realized by the choice of clinical outcome measure and weighting scheme. Although further verification is needed, we replicate a previous finding that the CFC2 may outperform the CDR-SB in the early AD population.

2026-04-14·ChemMedChem

Benzimidazole as a Versatile Scaffold for Developing Neurotherapeutics Against Neurodegenerative Diseases

Review

作者: Tiwari, Prashant ; Dubey, Sonal ; Barnwal, Nivedita

Neurodegenerative diseases (NDs) such as Alzheimer's disease (AD), Parkinson's disease (PD), Huntington's disease (HD), and amyotrophic lateral sclerosis (ALS) are characterized by progressive neuronal loss, leading to severe cognitive and motor dysfunction. Benzimidazole, a privileged heterocyclic scaffold, has emerged as a promising pharmacophore in modulating key pathological targets across these disorders. In AD, benzimidazole derivatives inhibit cholinesterases, glycogen synthase kinase‐3β (GSK‐3β), and glutaminyl cyclase (QC), thereby addressing cholinergic dysfunction, tau phosphorylation, and amyloid aggregation. In PD and HD, they act as monoamine oxidase‐B (MAO‐B) inhibitors, dopamine D1/D2 receptor modulators, and N‐methyl D‐aspartate receptor antagonists, improving dopaminergic signalling and reducing excitotoxicity. In ALS, benzimidazoles regulate acetylcholine dysfunction and inhibit receptor‐interacting protein kinase 1 (RIPK1), limiting neuroinflammation and cell death. Preclinical studies demonstrate potent enzyme inhibition, often with IC 50 values in the nanomolar to micromolar range, alongside favourable ADMET properties enabling blood‐brain barrier penetration. Clinically, the glutaminyl cyclase inhibitor Varoglutamstat has advanced to Phase II trials for AD, while Riluzole remains the only food and drug administration (FDA)‐approved benzimidazole drug for ALS. The structural versatility of benzimidazoles supports their development as multi‐target‐directed ligands, addressing overlapping mechanisms such as protein aggregation, oxidative stress, and neuroinflammation. Emerging strategies including hybrid molecules, nanocarrier delivery, and AI‐driven design may accelerate their clinical translation.

2025-12-01·Alzheimers & Dementia

Generalizability of digital speech and language assessments in MCI/mild AD across four non‐English languages

Article

作者: Schaeffer, Michael ; Xu, Mengdan ; Cormack, Francesca K ; Wenzkowski, Christine ; Lillo, Martina De ; Fuchs, Katharina ; Spilka, Michael J. ; Robin, Jessica

Abstract:

Background:

Neurodegenerative diseases affect speech production. Previous studies show various acoustic and linguistic changes in Mild Cognitive Impairment (MCI) and Alzheimer's disease (AD) speech patterns, such as more pauses and lower usage of nouns. However, most research has focused on English speakers. This analysis examines speech features in MCI and AD patients across four languages to determine if digital speech assessments yield comparable results across different languages.

Method:

This study included 244 patients with MCI or mild AD enrolled in the Phase 2b VIVIAD trial of varoglutamstat, in four language groups: Danish ( n =84), German ( n =59), Dutch ( n =40), and Spanish ( n =61). A picture description task was administered to all patients at baseline using instructions in their respective languages. Five speech features including number of picture objects described, number of nouns, speech rate, pronoun‐to‐noun ratio and average of word length were chosen based on previous analyses of speech patterns in MCI/AD, and a composite score was calculated combining the five measures. We tested the interrelationships among the features in each language and examined whether the speech features were related to another measure of cognitive function, the Mini‐Mental State Exam (MMSE). Associations were examined using a correlational approach.

Result:

There was variation in the magnitudes of the correlations between speech features in the different languages, but the directions of the associations were generally consistent across the four languages in the study. In all four languages, significant correlations with the MMSE were found with both the composite score (‐0.46 ≤ r ≤ ‐0.26) and the object content score (0.27 ≤ r ≤ 0.42).

Conclusion:

Together, these results indicate that there are consistent interrelationships among different speech features across languages. In addition, in the four languages examined, the speech composite score and the number of objects correctly described related to a standard measure of cognition, the MMSE. These results suggest that digital measures of speech and language in MCI/AD are generalizable across languages and relate to cognitive ability more generally. In addition, a composite score of multiple speech features may be more sensitive to cognitive impairment than individual speech metrics.

107

项与 Varoglutamstat 相关的新闻（医药）

2026-07-28

·精准药物

【Nucleic. Acids. Res】雷金平/杨跃东/王桢等报道概率扩散驱动的AI药物设计新思路

一、背景介绍先导化合物优化连接着早期命中与临床候选，是药物发现中依赖结构洞察、化学经验与反复试验的关键环节。传统配体生成模型多从已知分子的二维表示或配体分布中学习，难以充分利用靶标口袋的三维几何、局部化学环境以及决定药效的关键残基信息。因此，生成一个“化学上成立”的分子并不等同于生成一个真正适配靶标、具有高亲和力且能形成预期相互作用的候选药物。围绕这一核心问题，中山大学药学院雷金平副教授联合中山大学计算机学院杨跃东及王桢教授团队，以去噪概率扩散模型为统一技术底座，构建了由DiffDec、Diffleop和DiffDeCIG组成的分子设计与优化模型体系（图1）。首先，团队开发了靶标口袋三维结构感知的骨架修饰模型 DiffDec，让模型“看见口袋”，生成与口袋几何形状互补、化学环境匹配的分子；在此基础上，团队引入化学键约束和蛋白-配体亲和力引导，开发了 Diffleop，让模型“理解成键与亲和力”，将分子生成推向结构更合理、结合能力更强的化学空间；进一步地，团队通过引入氨基酸序列保守信息与蛋白-配体相互作用先验，开发了 DiffDeCIG，让模型“识别值得结合的残基”，引导生成的分子优先与功能关键的高保守位点形成相互作用。三个模型从靶标结构感知到亲和力引导再到机制导向设计，形成了一条完整的技术演进路线。图1. 靶向性质驱动的扩散分子优化模型的递进式技术路线二、文章亮点 2.1 DiffDec：让扩散模型在三维口袋中完成靶标结构感知的骨架修饰系列工作的第一步，是解决“生成基团能否真正适配靶标口袋”的问题。DiffDec （图2）将蛋白口袋与保留骨架作为固定条件，待生长的 R 基团原子坐标和类别在前向过程中逐步加噪，再由 E(3) 等变图神经网络在反向过程中恢复。等变结构保证模型对三维空间中的旋转和平移保持一致响应，从而能够学习口袋形状、化学环境与配体局部构象之间的对应关系。为实现不同尺寸 R 基团的端到端生成，DiffDec 进一步设计了“虚拟原子”机制：以统一长度的原子集合承载可变规模的基团，并在去噪后移除被判定为虚拟的节点。模型既可依据用户指定的锚点进行单点或多点修饰，也能自动识别合适的锚点位置。由此，分子骨架修饰不再是脱离靶标环境的二维片段拼接，而成为受靶标口袋几何和局部化学环境共同约束的三维生成问题。基准测试表明，DiffDec 在单点和多点骨架修饰任务中均表现出优异的生成能力。其中，单位点和多位点R基团恢复率分别达到 69.67% 和 45.34%，显著优于LibINVENT、Pocket2Mol和Flag等自回归模型。进一步的分子对接与构象分析显示，DiffDec 生成的分子不仅能够更好地占据靶标口袋，还可形成更有利的蛋白-配体相互作用，其键角和二面角分布也更接近真实分子。上述结果表明，将靶标口袋的三维结构约束直接纳入扩散生成过程，能够有效提升分子优化的准确性、结合潜力和构象合理性，为后续引入更明确的性质引导奠定了基础。图2. 靶标结构感知的DiffDec 模型框架 2.2 Diffleop：从口袋适配走向亲和力优化尽管三维口袋条件能够引导分子适配靶标环境，但单纯学习口袋与配体的联合分布，并不能为提升结合亲和力提供直接的驱动力；与此同时，仅对配体原子进行扩散，也难以充分保证生成分子的成键合理性。针对这些问题，团队在靶标结构感知的基础上进一步引入化学键约束和蛋白-配体亲和力引导，开发了Diffleop模型（图3），使分子生成同时面向“结合得更强”与“结构更合理”两个目标。 Diffleop在全连接分子图上对配体原子坐标、原子类型和化学键类型进行同步扩散，并以“虚拟键”表示原本不存在的连接，使模型能够在统一的去噪过程中学习化学键的分布。与此同时，团队构建了 E(3) 等变亲和力预测网络，在每一步采样中计算亲和力相对于原子坐标、原子类型和化学键类型的梯度，并利用该梯度引导分子生成朝着亲和力提升的方向进行。这样一来，模型不再只是学习“什么样的分子可能存在”，而是在生成过程中进一步判断“怎样生成才能结合得更强”。基准测试表明，Diffleop生成的分子在结合亲和力、蛋白-配体相互作用和配体效率等方面整体优于Pocket2Mol、AR、GraphBP、DiffLinker等模型，同时保持了合理的成键模式、局部几何构象和类药性质。消融实验进一步验证了两项引导信息的互补作用：亲和力引导决定分子向更强结合方向优化，化学键扩散则帮助模型维持结构合理性和类药性。研究团队随后跟四川大学李国菠教授湿实验团队合作，将Diffleop用于帕金森治疗谷氨酰环化酶sQC抑制剂的优化，成功获得3个活性显著提升的候选化合物，其中最优候选物A1的 sQC 抑制活性（IC50 = 8 nM）提升了226倍、比临床试验药物PQ912高出3.5倍，很好地展示了 Diffleop模型在先导化合物优化中的应用潜力。图3. 亲和力驱动的Diffleop模型框架 2.3 DiffDeCIG：从亲和力优化走向机制引导对于抗感染药物在内的很多药物而言，分子能否与功能关键且进化上高度保守的残基形成稳定相互作用，可能影响其作用机制、靶标选择性和潜在耐药屏障。基于此，研究团队进一步将氨基酸序列保守信息和蛋白-配体相互作用先验引入扩散生成过程，开发了 DiffDeCIG模型，使“优先与哪些残基形成相互作用”成为可被模型学习和执行的目标。 DiffDeCIG 首先通过多序列比对计算靶标序列的保守评分，并将其编码为蛋白口袋的附加特征；同时引入预训练的蛋白-配体相互作用先验网络 IPNet，学习蛋白与配体之间的关键作用模式。在扩散过程中，模型不仅学习了 R 基团原子坐标和类型的分布，还学习由相互作用先验产生的位置偏移。分子生成时，残基保守信息与相互作用先验共同调整 R 基团的生成方向，使新基团优先与高保守残基形成氢键、卤键、π-π堆积、疏水或盐桥等作用。基准测试表明，DiffDeCIG 生成的分子能够与口袋中高保守残基形成更多非键相互作用，以提高结合潜力和生物活性。这说明与高保守氨基酸形成相互作用不再是生成后的筛选指标，而是被直接转化为模型可学习的条件信息，从而将分子生成由亲和力优化进一步推进到关键位点和特定相互作用导向的设计。团队进一步跟中山大学周晖皓教授湿实验团队合作，将模型应用于甲硫氨酸转运酶MetRS抑制剂的优化，并成功获得了高活性化合物 MRS-9：它对金黄色葡萄球菌 MetRS 的抑制活性 IC50 达到 0.09 μM，较初始化合物提升超过300倍，并表现出对细菌型 MetRS 的良好选择性。晶体结构研究表明，MRS-9 形成了独特的三位点占据模式，验证了保守位点与相互作用引导在提升分子靶向生物活性和选择性方面的应用潜力。图4. 保守位点与相互作用引导的DiffDeCIG模型框架三、总结与展望从 DiffDec 到 DiffLeop，再到 DiffDeCIG，这一系列工作展示了概率扩散用于药物分子设计与优化的清晰递进：DiffDec 以三维口袋为条件，解决“生成分子是否放得进、配得上”的结构适配问题；Diffleop加入化学键约束和亲和力梯度，进一步回答“怎样生成得更合理、结合得更强”的问题；DiffDeCIG 则叠加残基保守性和相互作用先验，开始解决“应当与哪些氨基酸结合可能更有效”的机制导向问题。模型条件由几何环境扩展到物理化学性质，再延伸至进化与功能信息，推动生成式AI药物设计从分布拟合走向多层知识约束下的目标导向优化。更重要的是，两项后续工作均实现了真实药物发现的优化任务：Diffleop获得了体外活性优于临床试验药物的 sQC 抑制剂，DiffDeCIG 则获得了高活性、高选择性的 MetRS 三位点抑制剂，展现出从模型开发到真实药物发现的实际应用价值。参考文献： (1)Su, Jingtian#; Qiao, Anjie#; Huang, Weifeng; Xu, Jingyi; Lu, Feihu; Zhang, Hao; Deng, Qirui; Zou, Jialin; Wang, Zhen*; Lei, Jinping*; Zhou, Huihao*. Discovery of a triple-site inhibitor targeting bacterial methionyl-tRNA synthetase through combined drug repurposing screening and generative AI-assisted optimization. Nucleic Acids Research, 2026, 54, gkag488. (2)Anjie Qiao#, Yuting Chen#, Junjie Xie, Weifeng Huang, Hao Zhang, Qirui Deng, Jiahua Rao, Ji Deng, Fanbo Meng, Zhen Wang, Mingyuan Xu, Hongming Chen, Jiancong Xie, Shuangjia Zheng*, Yuedong Yang*, Guo-Bo Li*, Jinping Lei*. A 3D pocket-aware lead optimization model with knowledge guidance and its application for discovery of new glutaminyl cyclase inhibitors. Brief. Bioinform., 2025, 26, bbaf345. (3)Junjie Xie, Sheng Chen, Jinping Lei,* Yuedong Yang.* DiffDec: Structure-Aware Scaffold Decoration with an End-to-End Diffusion Model. J. Chem. Inf. Model, 2024, 64, 2554-2564. 声明：发表/转载本文仅仅是出于传播信息的需要，并不意味着代表本公众号观点或证实其内容的真实性。据此内容作出的任何判断，后果自负。若有侵权，告知必删！长按关注本公众号粉丝群/投稿/授权/广告等请联系公众号助手觉得本文好看，请点这里↓

2026-07-22

·BioSpace

Vivoryon Therapeutics N.V. to Report H1 2026 Financial Results and Business Updates on August 6, 2026; Annual General Meeting Planned for End of September 2026

Vivoryon Therapeutics N.V. to Report H1 2026 Financial Results and Business Updates on August 6, 2026; Annual General Meeting Planned for End of September 2026 Halle (Saale) / Munich, Germany, July 22 , 2026 - Vivoryon Therapeutics N.V. (Euronext Amsterdam: VVY; NL00150002Q7) ( Vivoryon ), a clinical stage company developing small molecule medicines for inflammatory and fibrotic disorders, with a primary focus on kidney diseases, today announced that it will report half year financial results and business updates for the period ended June 30, 2026, on Thursday, August 6, 2026, and that it plans to host its 2026 Annual General Meeting at the end of September 2026. The convocation note and all related materials will be made available on the Company’s website in due time. For the H1 2026 financial results, the Company will host a conference call and webcast open to the public. The report will be available for download on the Company’s website ( ). H1 2026 conference call details Date: August 6, 2026 Time: 3:00 pm CEST / 9:00 am EDT The H1 2026 conference call will be available via phone and webcast. The live audio webcast of the call will be available on Vivoryon’s website at: To join the H1 2026 conference call via phone, participants may pre-register and will receive dedicated dial-in details to easily and quickly access the call via the following website: ezstandalone.cmd.push(function () { ezstandalone.showAds(131); }); It is suggested participants dial into the conference call 15 minutes prior to the scheduled start time to avoid any delays in attendance. ### About Vivoryon Therapeutics N.V. Vivoryon is a clinical stage biotechnology company focused on developing innovative small molecule-based medicines for the treatment of inflammatory and fibrotic disorders of the kidney. Driven by its passion for ground-breaking science and innovation, the Company strives to improve patient outcomes by changing the course of severe diseases through modulating the activity and stability of pathologically relevant proteins. Vivoryon’s most advanced program, varoglutamstat, a proprietary, first-in-class orally available QPCT/L inhibitor, is being evaluated to treat diabetic kidney disease. Vivoryon Forward Looking Statements This press release includes forward-looking statements, including, without limitation, those regarding the business strategy, management plans and objectives for future operations of Vivoryon Therapeutics N.V. (the “Company”), estimates and projections with respect to the market for the Company’s products and forecasts and statements as to when the Company’s products may be available. Words such as “anticipate,” “believe,” “estimate,” “expect,” “forecast,” “intend,” “may,” “plan,” “project,” “predict,” “should” and “will” and similar expressions as they relate to the Company are intended to identify such forward-looking statements. These forward-looking statements are not guarantees of future performance; rather they are based on the Management’s current expectations and assumptions about future events and trends, the economy and other future conditions. The forward-looking statements involve a number of known and unknown risks and uncertainties. These risks and uncertainties and other factors could materially adversely affect the outcome and financial effects of the plans and events described herein. The Company’s results of operations, cash needs, financial condition, liquidity, prospects, future transactions, strategies or events may differ materially from those expressed or implied in such forward-looking statements and from expectations. As a result, no undue reliance should be placed on such forward-looking statements. This press release does not contain risk factors. Certain risk factors that may affect the Company’s future financial results are discussed in the published annual financial statements of the Company. This press release, including any forward-looking statements, speaks only as of the date of this press release. The Company does not assume any obligation to update any information or forward-looking statements contained herein, save for any information required to be disclosed by law. For more information, please contact: Vivoryon Therapeutics N.V. Dr. Manuela Bader, Director IR & Communication Email: IR@vivoryon.com LifeSci Advisors Sandya von der Weid Tel: +41 78 680 05 38 Email: svonderweid@lifesciadvisors.com Attachment 20260722_H1 2026 NoR_

财报

2026-07-13

·DrugAI

Nucleic. Acids Res.| 概率扩散驱动的AI药物设计：从靶标结构感知到亲和力驱动再到保守相互作用引导

先导化合物优化连接着早期命中与临床候选，是药物发现中依赖结构洞察、化学经验与反复试验的关键环节。传统配体生成模型多从已知分子的二维表示或配体分布中学习，难以充分利用靶标口袋的三维几何、局部化学环境以及决定药效的关键残基信息。因此，生成一个“化学上成立”的分子并不等同于生成一个真正适配靶标、具有高亲和力且能形成预期相互作用的候选药物。围绕这一核心问题，中山大学药学院雷金平副教授联合中山大学计算机学院杨跃东及王桢教授团队，以去噪概率扩散模型为统一技术底座，构建了由DiffDec、Diffleop和DiffDeCIG组成的分子设计与优化模型体系（图1）。首先，团队开发了靶标口袋三维结构感知的骨架修饰模型 DiffDec，让模型“看见口袋”，生成与口袋几何形状互补、化学环境匹配的分子；在此基础上，团队引入化学键约束和蛋白-配体亲和力引导，开发了 Diffleop，让模型“理解成键与亲和力”，将分子生成推向结构更合理、结合能力更强的化学空间；进一步地，团队通过引入氨基酸序列保守信息与蛋白-配体相互作用先验，开发了 DiffDeCIG，让模型“识别值得结合的残基”，引导生成的分子优先与功能关键的高保守位点形成相互作用。三个模型从靶标结构感知到亲和力引导再到机制导向设计，形成了一条完整的技术演进路线。图1. 靶向性质驱动的扩散分子优化模型的递进式技术路线。文章亮点 DiffDec：让扩散模型在三维口袋中完成靶标结构感知的骨架修饰系列工作的第一步，是解决“生成基团能否真正适配靶标口袋”的问题。DiffDec （图2）将蛋白口袋与保留骨架作为固定条件，待生长的 R 基团原子坐标和类别在前向过程中逐步加噪，再由 E(3) 等变图神经网络在反向过程中恢复。等变结构保证模型对三维空间中的旋转和平移保持一致响应，从而能够学习口袋形状、化学环境与配体局部构象之间的对应关系。为实现不同尺寸 R 基团的端到端生成，DiffDec 进一步设计了“虚拟原子”机制：以统一长度的原子集合承载可变规模的基团，并在去噪后移除被判定为虚拟的节点。模型既可依据用户指定的锚点进行单点或多点修饰，也能自动识别合适的锚点位置。由此，分子骨架修饰不再是脱离靶标环境的二维片段拼接，而成为受靶标口袋几何和局部化学环境共同约束的三维生成问题。基准测试表明，DiffDec 在单点和多点骨架修饰任务中均表现出优异的生成能力。其中，单位点和多位点R基团恢复率分别达到 69.67% 和 45.34%，显著优于LibINVENT、Pocket2Mol和Flag等自回归模型。进一步的分子对接与构象分析显示，DiffDec 生成的分子不仅能够更好地占据靶标口袋，还可形成更有利的蛋白-配体相互作用，其键角和二面角分布也更接近真实分子。上述结果表明，将靶标口袋的三维结构约束直接纳入扩散生成过程，能够有效提升分子优化的准确性、结合潜力和构象合理性，为后续引入更明确的性质引导奠定了基础。图2. 靶标结构感知的DiffDec 模型框架。 Diffleop：从口袋适配走向亲和力优化尽管三维口袋条件能够引导分子适配靶标环境，但单纯学习口袋与配体的联合分布，并不能为提升结合亲和力提供直接的驱动力；与此同时，仅对配体原子进行扩散，也难以充分保证生成分子的成键合理性。针对这些问题，团队在靶标结构感知的基础上进一步引入化学键约束和蛋白-配体亲和力引导，开发了Diffleop模型（图3），使分子生成同时面向“结合得更强”与“结构更合理”两个目标。 Diffleop在全连接分子图上对配体原子坐标、原子类型和化学键类型进行同步扩散，并以“虚拟键”表示原本不存在的连接，使模型能够在统一的去噪过程中学习化学键的分布。与此同时，团队构建了 E(3) 等变亲和力预测网络，在每一步采样中计算亲和力相对于原子坐标、原子类型和化学键类型的梯度，并利用该梯度引导分子生成朝着亲和力提升的方向进行。这样一来，模型不再只是学习“什么样的分子可能存在”，而是在生成过程中进一步判断“怎样生成才能结合得更强”。基准测试表明，Diffleop生成的分子在结合亲和力、蛋白-配体相互作用和配体效率等方面整体优于Pocket2Mol、AR、GraphBP、DiffLinker等模型，同时保持了合理的成键模式、局部几何构象和类药性质。消融实验进一步验证了两项引导信息的互补作用：亲和力引导决定分子向更强结合方向优化，化学键扩散则帮助模型维持结构合理性和类药性。研究团队随后跟四川大学李国菠教授湿实验团队合作，将Diffleop用于帕金森治疗谷氨酰环化酶sQC抑制剂的优化，成功获得3个活性显著提升的候选化合物，其中最优候选物A1的 sQC 抑制活性（IC50 = 8 nM）提升了226倍、比临床试验药物PQ912高出3.5倍，很好地展示了 Diffleop模型在先导化合物优化中的应用潜力。图3. 亲和力驱动的Diffleop模型框架。 DiffDeCIG：从亲和力优化走向机制引导对于抗感染药物在内的很多药物而言，分子能否与功能关键且进化上高度保守的残基形成稳定相互作用，可能影响其作用机制、靶标选择性和潜在耐药屏障。基于此，研究团队进一步将氨基酸序列保守信息和蛋白-配体相互作用先验引入扩散生成过程，开发了 DiffDeCIG模型，使“优先与哪些残基形成相互作用”成为可被模型学习和执行的目标。 DiffDeCIG 首先通过多序列比对计算靶标序列的保守评分，并将其编码为蛋白口袋的附加特征；同时引入预训练的蛋白-配体相互作用先验网络 IPNet，学习蛋白与配体之间的关键作用模式。在扩散过程中，模型不仅学习了 R 基团原子坐标和类型的分布，还学习由相互作用先验产生的位置偏移。分子生成时，残基保守信息与相互作用先验共同调整 R 基团的生成方向，使新基团优先与高保守残基形成氢键、卤键、π-π堆积、疏水或盐桥等作用。基准测试表明，DiffDeCIG 生成的分子能够与口袋中高保守残基形成更多非键相互作用，以提高结合潜力和生物活性。这说明与高保守氨基酸形成相互作用不再是生成后的筛选指标，而是被直接转化为模型可学习的条件信息，从而将分子生成由亲和力优化进一步推进到关键位点和特定相互作用导向的设计。团队进一步跟中山大学周晖皓教授湿实验团队合作，将模型应用于甲硫氨酸转运酶MetRS抑制剂的优化，并成功获得了高活性化合物 MRS-9：它对金黄色葡萄球菌 MetRS 的抑制活性 IC50 达到 0.09 μM，较初始化合物提升超过300倍，并表现出对细菌型 MetRS 的良好选择性。晶体结构研究表明，MRS-9 形成了独特的三位点占据模式，验证了保守位点与相互作用引导在提升分子靶向生物活性和选择性方面的应用潜力。图4. 保守位点与相互作用引导的DiffDeCIG模型框架。总结与展望从 DiffDec 到 DiffLeop，再到 DiffDeCIG，这一系列工作展示了概率扩散用于药物分子设计与优化的清晰递进：DiffDec 以三维口袋为条件，解决“生成分子是否放得进、配得上”的结构适配问题；Diffleop加入化学键约束和亲和力梯度，进一步回答“怎样生成得更合理、结合得更强”的问题；DiffDeCIG 则叠加残基保守性和相互作用先验，开始解决“应当与哪些氨基酸结合可能更有效”的机制导向问题。模型条件由几何环境扩展到物理化学性质，再延伸至进化与功能信息，推动生成式AI药物设计从分布拟合走向多层知识约束下的目标导向优化。更重要的是，两项后续工作均实现了真实药物发现的优化任务：Diffleop获得了体外活性优于临床试验药物的 sQC 抑制剂，DiffDeCIG 则获得了高活性、高选择性的 MetRS 三位点抑制剂，展现出从模型开发到真实药物发现的实际应用价值。参考资料 Su, Jingtian#; Qiao, Anjie#; Huang, Weifeng; Xu, Jingyi; Lu, Feihu; Zhang, Hao; Deng, Qirui; Zou, Jialin; Wang, Zhen*; Lei, Jinping*; Zhou, Huihao*. Discovery of a triple-site inhibitor targeting bacterial methionyl-tRNA synthetase through combined drug repurposing screening and generative AI-assisted optimization. Nucleic Acids Research, 2026, 54, gkag488. Anjie Qiao#, Yuting Chen#, Junjie Xie, Weifeng Huang, Hao Zhang, Qirui Deng, Jiahua Rao, Ji Deng, Fanbo Meng, Zhen Wang, Mingyuan Xu, Hongming Chen, Jiancong Xie, Shuangjia Zheng*, Yuedong Yang*, Guo-Bo Li*, Jinping Lei*. A 3D pocket-aware lead optimization model with knowledge guidance and its application for discovery of new glutaminyl cyclase inhibitors. Brief. Bioinform., 2025, 26, bbaf345. Junjie Xie, Sheng Chen, Jinping Lei,* Yuedong Yang.* DiffDec: Structure-Aware Scaffold Decoration with an End-to-End Diffusion Model. J. Chem. Inf. Model, 2024, 64, 2554-2564.

100 项与 Varoglutamstat 相关的药物交易

登录后查看更多信息

研发状态

10 条进展最快的记录,

后查看更多信息

适应症	最高研发状态	国家/地区	公司	日期
轻度痴呆	临床2期	中国	Haupt Pharma Wülfing GmbH	2022-04-26
轻度痴呆	临床2期	中国	江苏先声药业有限公司	2022-04-26
轻度痴呆	临床2期	中国	Vivoryon Therapeutics NV	2022-04-26
轻度认知障碍	临床2期	丹麦	Vivoryon Therapeutics NV	2020-07-06
轻度认知障碍	临床2期	丹麦	Nordic Bioscience A/S	2020-07-06
轻度认知障碍	临床2期	德国	Nordic Bioscience A/S	2020-07-06
轻度认知障碍	临床2期	德国	Vivoryon Therapeutics NV	2020-07-06
轻度认知障碍	临床2期	荷兰	Nordic Bioscience A/S	2020-07-06
轻度认知障碍	临床2期	荷兰	Vivoryon Therapeutics NV	2020-07-06
轻度认知障碍	临床2期	波兰	Vivoryon Therapeutics NV	2020-07-06

登录后查看更多信息

临床结果

适应症

分期

评价

查看全部结果

研究	分期	人群特征	评价人数	分组	结果	评价	发布日期


NCT03919162 (VIVA-MIND, CTgov)	临床2期	阿尔茨海默症	112	PQ912 (PQ912)	獵淵觸積艱鏇製構鹹糧 = 顧顧觸遞簾鏇範糧鬱構簾鬱構壓蓋願襯簾醖製 (醖鹹鹹壓膚襯製蓋鏇醖, 簾願餘獵選觸壓網構餘 ~ 選襯廠襯觸淵憲膚膚襯) 更多	-	2025-12-12
				Placebo (Placebo)	獵淵觸積艱鏇製構鹹糧 = 繭鑰醖壓鹽鑰獵觸淵鑰簾鬱構壓蓋願襯簾醖製 (醖鹹鹹壓膚襯製蓋鏇醖, 觸繭繭窪夢糧繭構鬱顧 ~ 餘網顧鑰窪齋繭鹹鏇網) 更多
NCT03919162 (VIVA-MIND, GlobeNewswire) 人工标引	临床2期	阿尔茨海默症	-	Varoglutamstat	遞壓鬱鑰選憲艱築衊繭(簾廠鹽構襯夢衊餘遞襯) = no clinically meaningful and no statistically significant differences between varoglutamstat 600mg BID and placebo 夢艱壓膚齋齋夢繭積積 (糧願淵糧構醖淵醖顧襯 ) 更多	积极	2024-12-09
				placebo
NCT04498650 (VIVIAD, ASN2024)	临床2期	阿尔茨海默症 \| 轻度认知障碍	259	Varoglutamstat 300 mg	糧膚鏇製糧製窪鹽獵鹹(鹹鹹鹹壓遞窪製繭鬱鑰) = 餘醖憲願鬱簾膚範窪願衊壓廠繭窪襯壓憲齋壓 (鹽構鏇鏇範鏇選鏇廠齋 )	积极	2024-10-25
				Varoglutamstat 600 mg	廠積範夢築襯壓齋觸鏇(範糧壓鏇鬱糧壓鏇鑰觸) = 淵範鹹淵膚網壓遞壓膚壓繭窪鹹簾餘鹽範壓窪 (廠夢鏇繭積夢鏇鬱選餘 ) 更多
NCT04498650 (VIVIAD, GlobeNewswire \| ASN2024)	临床2期	阿尔茨海默症 \| 糖尿病肾病	32	Varoglutamstat	網膚積鏇醖鹹選鏇遞觸(網遞鬱繭選鹹觸繭築淵) = 糧鏇網糧襯衊繭醖網膚築選夢廠鑰醖齋鹹壓糧 (觸簾夢廠夢艱製醖繭艱 )	积极	2024-10-25
				placebo	網膚積鏇醖鹹選鏇遞觸(網遞鬱繭選鹹觸繭築淵) = 憲廠襯糧觸顧窪鹽顧鏇築選夢廠鑰醖齋鹹壓糧 (觸簾夢廠夢艱製醖繭艱 )
NCT04498650 (VIVIAD, GlobeNewswire) 人工标引	临床2期	肾脏疾病 CCL2 Expression	-	Varoglutamstat 600mg BID	醖鹹鹹襯衊衊獵醖鬱餘(鏇積廠選艱壓鏇顧壓遞) = 範網蓋餘糧餘餘壓廠遞廠構鹽製繭糧簾廠窪夢 (夢窪鹹鹽簾鬱簾觸鹹願 ) 更多	积极	2024-09-12
NCT04498650 (VIVIAD, Corporate Publications) 人工标引	临床2期	阿尔茨海默症	250	Varoglutamstat	積鏇積憲顧鏇選鑰繭鹹(糧選膚壓鏇構襯夢網醖) = 窪積餘選鏇鑰糧襯憲觸憲製網獵窪糧齋網獵膚 (遞淵醖構壓鑰膚觸網窪 ) 更多	积极	2022-07-31
NCT02389413 (Pubmed \| Alzheimers Res Ther) 人工标引	临床2期	阿尔茨海默症	120	PQ912	餘鑰繭夢窪積餘顧廠鑰(齋願壓齋餘衊醖蓋觸觸) = There was no significant difference between treatments in the number of subjects with (serious) adverse events, although there were slightly more patients with a serious adverse event in the PQ912 group compared to placebo. 蓋廠膚淵獵鏇構糧觸襯 (簾壓選獵壓鏇構廠積膚 ) 更多	积极	2018-10-12
				Placebo