股票 
板块 
用户 
资讯 Swipe Left for English News→
摘要
在制药工艺开发中，标准优化工具对于提高反应收率和通量具有不可替代的价值。然而，目前传统优化方法遇到了瓶颈，它们无法解释反应为何出现不可预测的行为。本白皮书将探讨如何通过将反应动力学、光谱学和计算化学等先进的机理研究方法有机结合，来降低生产风险、解析意外的工艺偏差，甚至是发掘全新的、有价值的反应路径。
通过对酰胺还原中催化剂负载量依赖的化学选择性的案例研究，我们展示了基于第一性原理的化学理解如何为建立稳健的、端到端的商业化生产控制策略提供必要的科学基础。
引言：经验优化工具的局限性
在合同研发生产组织（CDMO）中，工艺开发通常依赖于旨在快速提高收率、选择性、稳健性和可放大性的实用的优化工具包。尽管这些工具极为有效，但经验性筛选工具本质上侧重于相关性，而非因果关系。
研究方法
核心优势
实际局限性
OVAT
（单变量法）
直观、易于理解；通过改变单一变量（温度、投料量、化学计量比）即可快速识别方向性趋势。
速度较慢，且会遗漏多变量交互作用。例如，由于高温下竞争性副反应的存在，最佳试剂当量数可能高度依赖于温度。
贝叶斯优化
数据高效、模型指导的方法，通过序贯选择实验以最大化目标指标（收率、纯度、通量）。非常适合于物料有限、设计空间大的情况。
所获取的信息高度依赖于初始模型空间；若提前达到目标指标，则可能遗漏关键边界条件的依赖性。
DoE
（实验设计法）
结构化映射设计空间，以最小的实验量获得统计显著的交互作用数据。
能够识别具有统计显著性的因素与预测操作边界，无法识别意外化学偏差的根本原因。
图1. 经验优化工具包对比
这些方法是互补的，OVAT（单变量法）、贝叶斯优化和 DoE（实验设计）有助于确定如何改进反应、哪些因素最为关键，以及在限定条件下预期获得何种结果。因此，它们是 J-STAR 催化与赋能技术团队不可或缺的工具。
然而，即使这些相辅相成的方法能成功优化各项工艺，它们仍然将反应视为一个“黑盒”，缺乏分子水平的理解。隐藏变量、瞬态杂质、竞争性路径或依赖于规模效应的混合效应可能始终未被识别，从而在后期开发或商业化生产放大阶段导致灾难性的、始料未及的后果。
机理研究在风险管控中的核心价值
机理研究通过将宏观实验观察与基于第一性原理的化学理解直接关联，从根本上填补了这一认知缺口。借助机理研究的系统化方法，工艺科学家不必再靠猜测来解释公斤级放大中收率骤降的原因，而是可以通过机理研究工作流程实现以下目标：
厘清催化剂物种形态，区分活性与非活性催化形式。
识别瞬态中间体，并捕捉短寿命的降解路径。
找准真正的限速步骤，以合理优化循环时间。
揭示催化剂抑制与失活路径，明确痕量杂质、溶剂稳定剂或反应器材质的影响。
区分有效反应路径与无效反应路径，以设计坚实可靠的控制策略。
长期以来，机理研究在 CDMO 领域未得到充分利用，这是因为它们需要专业的知识储备、专用的分析设备以及紧密的跨学科协作。
通过无缝整合多学科技术手段——以光谱技术（NMR/IR）检测反应中间体、以动力学分析建立反应规律的数学模型、以计算化学（DFT）评估过渡态——客户的注册申报和控制策略便能获得扎实的科学支撑。
案例研究：催化酰胺还原中的发散反应性与创新
博腾 J-STAR 工艺化学与赋能技术团队长期致力于通过综合机理研究解决工艺开发中的关键挑战。团队针对某反应在低催化剂负载下转化不完全的问题，结合机理分析找到了根源，并在此过程中意外发现了一个全新的催化反应，可用于高价值砌块分子的选择性合成[1]。 这是博腾 J-STAR 团队工艺化学与赋能技术紧密协作、协同创新的众多范例之一。
烯胺是有机合成中重要的砌块分子，可参与多种转化，包括区域选择性烷基化、酰基化、串联环化反应、环加成反应以及杂环形成反应。它们还可通过还原直接转化为相应的胺，这也是药物合成中的一类重要转化。烯胺最常用的制备方法是在路易斯酸或布朗斯特酸催化下，由仲胺与羰基化合物缩合而成。其他方法包括过渡金属催化乙烯基卤化物与胺的交叉偶联、烷基胺脱氢以及炔烃的氢胺化反应。尽管取得了这些进展，但许多方法需要苛刻的反应条件，并且很大程度上局限于酮衍生烯胺。醛衍生烯胺的制备仍然较为困难，通常需要强路易斯酸、高温、化学计量的脱水剂或较长的反应时间，这限制了这些方法的实际应用范围。
在此研究中，团队证明了单一催化体系可通过简单的催化剂负载量调节，实现部分还原与完全还原酰胺之间的切换。传统方法中，选择性酰胺还原需要改变催化剂结构、添加剂或更广泛反应条件；而研发团队在此提供了一种不同的策略，即烯胺生成的化学选择性仅由催化剂负载量控制。以 Ru3(CO)12 为催化剂时，低催化剂负载量显著倾向于生成烯胺，而较高负载量则促进完全还原为胺。烯胺一旦生成，即使在能将起始酰胺直接还原为胺的条件下，也表现出良好的化学惰性，不会发生过度还原。这一结果表明，烯胺并非完全还原成胺路径上的简单中间体。
研发团队认为，催化剂负载量依赖的反应性差异反映了同一体系内两种不同的催化机制（图2）。实验与计算研究表明，烯胺和胺产物均源自共同的亚胺离子中间体。但烯胺并不直接作为胺的前体，而是作为主催化循环之外的产物存在。
图2. 烯胺与亚胺的还原
DFT 计算进一步评估了烯胺与亚胺的相对稳定性（图3）、E/Z 烯胺选择性的形成机制以及可能的通过亚胺离子中间体发生的异构化过程，这对理解反应机制具有深远意义。当亚胺离子形成后，低催化剂负载量会导致较低的稳态亚胺正离子浓度与有限的催化剂催化转化能力，在动力学上抑制进一步的还原，从而实现了稳定的烯胺生成。相反，较高的催化剂负载量则促进亚胺中间体的还原，使胺成为主要产物。附加实验表明，弱酸性物种（如 TMSOH）可将烯胺转化为其亚胺互变异构体，后者随后可被进一步还原为胺。
图3. 可能的反应机理
酰胺选择性还原为烯胺的反应在广泛的底物范围内均表现出高选择性（图4）。
图4. 酰胺转化为烯胺的底物范围
结论
通过将先进的机理研究方法（如反应动力学、光谱学与计算化学）与传统的工艺优化工具深度整合，工艺研发人员能够超越经验性的“黑盒”优化，获得对化学反应本质的透彻理解。正如酰胺还原案例所示，这种基于第一性原理的策略不仅能有效识别并降低放大生产中的风险，解析意料之外的工艺偏差，更能开启发现全新、高价值反应路径的大门。这对于建立稳健、可重复且符合监管要求的商业化生产控制策略，具有不可替代的战略意义。
博腾赋能技术能力介绍
开展化学工艺放大生产时，能否找到一家脱离反复试错、依托科学机理开展研发的合作伙伴，是决定验证批次顺利落地，还是出现成本高昂的生产失败的关键因素。博腾 J-STAR 团队致力于把反应机理的深入理解融入到每一个项目里，将工艺化学、分析支持与赋能技术整合在同一平台，帮助客户在缩短 CMC 周期的同时，对关键工艺参数形成更系统、更符合监管要求的认知，从源头降低项目风险。
在三大先进技术赋能平台上，博腾持续投入研发与创新：
金属催化：高通量筛选结合经验优化与机理研究，攻克复杂的偶联反应与不对称还原。
生物催化：定制化酶工程与筛选，以实现高度绿色、高化学选择性的转化。
连续反应：通过连续流工艺安全处理高能试剂、极端温度、高压反应及光化学反应。
通过将先进技术平台与深刻严谨的机理分析相结合，博腾将持续为客户提供稳健、可放大且可靠的 CMC 解决方案，突破制造瓶颈，与客户携手让好药更早惠及大众。
注：
1.此工作已作为研究论文发表，标题为《Ru-Catalyzed Reductive Dehydration of Amides to Enamines: Catalyst-Loading-Controlled Selectivity》，作者为Jacob Y. Cha*博士及博腾团队（Ziqing Zuo, William B. Reid, Todd Nelson 及 Thorsten Rosner）；ACS Catal. 2026, 16, 4985?4992. https://doi.org/10.1021/acscatal.5c09022
END
博腾股份成立于2005年，为全球药企、生物科技公司及科研机构等提供从临床前研究到药品上市全生命周期的端到端CDMO服务，涵盖小分子药物、多肽与寡核苷酸药物、蛋白与偶联药物以及细胞与基因治疗药物的原料药到制剂全流程。截至2025年末，公司拥有全球员工4300余人，累计服务客户1400余家，成功交付项目超5000个。我们始终坚持以客户为中心，致力于提供创新可靠的全球化CDMO解决方案，让好药更早惠及大众。
免责声明：本文内容仅供参考，旨在分享相关技术、案例及行业观点，不构成任何药物研发或其他专业建议。博腾及其子公司不保证本文内容的准确性、完整性或时效性，也不对因使用或依赖本文内容而产生的任何责任负责。读者应结合最新科学证据、适用法规及专业意见进行独立评估。
Executive Summary
In pharmaceutical process development, standard optimization tools are invaluable for improving yields and throughput. However, when conventional methods hit a wall, they cannot explain why a reaction behaves unpredictably. This white paper explores how integrating advanced mechanistic studies—combining reaction kinetics, spectroscopy, and computational chemistry—mitigates manufacturing risks, explains unexpected process deviations, and can even uncover entirely new, valuable reaction pathways.
Through a case study on loading-dependent chemoselectivity in amide reductions, we demonstrate how first-principles chemical understanding provides the scientific foundation necessary for robust, end-to-end commercial manufacturing control strategies.
Introduction: The Limitations of Empirical Optimization
Process development at a Contract Development and Manufacturing Organization (CDMO) typically relies on practical optimization toolkits designed to rapidly improve yield, selectivity, robustness, and scalability. While highly effective, empirical screening tools inherently focus on correlation rather than causation.
Methodology
Core Strength
Practical Limitations
OVAT
(One-Variable-at-a-Time)
Intuitive, easy to interpret; quickly identifies directional trends by shifting a single factor (temperature, loading, stoichiometry).
Can be slow and misses multi-variable interactions. For example, optimal reagent equivalents may heavily depend on temperature due to competing side-reactions at elevated temperatures.
Bayesian Optimization
Data-efficient, model-guided approach that sequentially selects experiments to maximize a target objective (yield, purity, throughput). Ideal for large design spaces with limited material.
Information gained is highly dependent on the initial model space; if the target objective is reached early, critical edge-case dependencies may remain unexplored.
DoE
(Design of Experiments)
Structured mapping of the design space to minimize experimental burden while extracting statistically meaningful data on interactions.
Identifies statistically significant factors and predictive operating boundaries but cannot identify the underlying root cause of an unexpected chemical deviation.
Figure 1. Empirical optimization toolkit comparison
These approaches are complementary. OVAT, Bayesian optimization, and DoE help determine how to improve a reaction, which factors matter most, and what outcome is expected under defined conditions. They are therefore essential tools for the Catalysis and Enabling Technology group at J-STAR.
Ultimately, even when these complementary methods successfully optimize a process, they treat the reaction as a "black box". Without a molecular-level understanding, hidden variables, transient impurities, competing pathways, or scale-dependent mixing effects can remain entirely unrecognized—leading to catastrophic, unexpected results during late-stage development or commercial manufacturing scale-up.
The Power of Mechanistic Insight in Risk Mitigation
Mechanistic studies address this fundamental knowledge gap by connecting macro-scale experimental observations directly to first-principles chemical understanding. Rather than guessing why a yield suddenly drops on a multi-kilogram scale, mechanistic workflows allow process scientists to:
Clarify catalyst speciation and track active vs. inactive catalytic forms.
Identify transient intermediates and isolate short-lived degradation pathways.
Determine true rate-limiting steps to optimize cycle times rationally.
Reveal catalyst inhibition or deactivation pathways caused by trace impurities, solvent stabilizers, or reactor metallurgy.
Distinguish productive from nonproductive reaction pathways to design bulletproof control strategies.
Historically, mechanistic studies have been underutilized in CDMO settings because they demand specialized expertise, fit-for-purpose analytical instrumentation, and tight cross-disciplinary collaboration.
By seamlessly integrating spectroscopy (NMR/IR) to detect reactive intermediates, kinetic analysis to mathematically model reaction dependencies, and computational chemistry (DFT) to evaluate transition states, clients gain solid scientific justification for their regulatory filings and control strategies.
Case Study: Divergent Reactivity and Innovation in Catalytic Amide Reductions
The following case study illustrates how an integrated mechanistic approach successfully resolved an incomplete conversion issue at lower catalyst loadings and, in doing so, uncovered an entirely new catalytic pathway for the selective synthesis of valuable building blocks.[1] This work is one of many successful examples of a strong collaborative effort between our Process Chemistry and Enabling Technology Group.
Enamines are valuable building blocks in organic synthesis and participate in many transformations, including regioselective alkylations, acylations, annulation cascades, cycloadditions, and heterocycle-forming reactions. They can also be converted directly to the corresponding amines by reduction, an important transformation in pharmaceutical synthesis. Enamines are most commonly prepared by condensation of a secondary amine with a carbonyl compound under Lewis or Br?nsted acid catalysis. Other approaches include transition-metal-catalyzed cross-coupling of vinyl halides with amines, dehydrogenation of alkyl amines, and hydroamination of alkynes. Despite these advances, many methods require harsh conditions and are largely limited to ketone-derived enamines. Aldehyde-derived enamines remain less accessible, often requiring strong Lewis acids, elevated temperatures, stoichiometric dehydrating reagents, or long reaction times, which narrow their practical scope.
In this work, the team showed that a single catalytic system can be tuned to switch between partial and complete amide reduction. Traditionally, selective amide reduction requires changes to the catalyst structure, additives, or broader reaction conditions. Here, the authors reported a different strategy in which chemoselectivity for enamine formation is controlled solely by catalyst loading. Using Ru3(CO)12 as the catalyst, low loadings strongly favored enamine formation, whereas higher loadings promoted complete reduction to the amine. Importantly, once formed, the enamine remained chemically stable and avoided overreduction, even under conditions that could reduce the starting amide directly to the amine. This result demonstrated that the enamine was not simply an intermediate on the pathway to the fully reduced amine.
The team proposed that the loading-dependent divergence in reactivity reflects two distinct catalytic regimes within the same system (Figure 2). Experimental and computational studies indicated that both enamine and amine products arise from a common iminium ion intermediate. However, the enamine does not act as a direct intermediate enroute to the amine; instead, it represents an off-cycle product.
Figure 2. Reductions of Enamine and Imine
DFT calculations evaluated the relative stabilities of enamines and imines (Figure 3), the origin of E/Z enamine selectivity, and possible isomerization through an iminium intermediate, providing important mechanistic insight. Once the iminium ion forms, low catalyst loading leads to a low steady-state iminium concentration and limited catalytic turnover, which kinetically suppresses further reduction and enables stable enamine formation. Higher catalyst loading instead promotes reduction of the iminium intermediate, giving the amine as the major product. Additional experiments showed that mildly acidic species, such as TMSOH, can shift the enamine toward its iminium tautomer, which can then undergo further reduction to the amine.
Figure 3. Proposed Reaction Mechanism
The selective reduction of amides to enamines was demonstrated across a broad substrate scope with high selectivity (Figure 4).
Figure 4. Substrate Scope of Amides to Enamines
Conclusion
By deeply integrating advanced mechanistic studies—such as reaction kinetics, spectroscopy, and computational chemistry—with traditional process optimization tools, process chemists can move beyond empirical “black box” operations and gain a thorough understanding of the fundamental nature of chemical reactions. As demonstrated in the amide reduction case study, this first-principles-based strategy not only enables effective identification and mitigation of scale-up risks and resolution of process deviations that conventional methods fail to explain, but also opens the door to the discovery of entirely new, high-value reaction pathways. This holds strategic significance for establishing robust, reproducible, and regulatory-compliant commercial manufacturing control strategies.
Porton’s Enabling Technology Capabilities
When scaling up advanced chemical processes, having a partner capable of moving beyond empirical trial-and-error is the difference between a successful validation campaign and a costly manufacturing failure. The Porton J-STAR team is committed to seamlessly integrating deep mechanistic insights into project workflows. By unifying process chemistry, analytical support, and enabling technologies under one roof, we accelerate our clients’ CMC timelines while building a deeper, regulatory-compliant understanding of their asset's critical process parameters, mitigating risks from the outset.
Porton continues to pioneer sustainable, scalable, and highly innovative chemical technologies across primary advanced platforms:
Clarify catalyst speciation and track active vs. inactive catalytic forms.
Metal Catalysis: High-throughput screening, empirical and mechanistic optimization of complex cross-couplings and asymmetric reductions.
Biocatalysis: Custom enzyme engineering and screening for highly green, chemoselective transformations.
Flow Chemistry: Continuous processing to safely handle energetic reagents, extreme temperatures, high-pressure reactions, and photochemical reactions.
By blending advanced technology platforms with deep mechanistic rigor, Porton delivers robust, scalable, and reliable CMC solutions that overcome tough manufacturing bottlenecks and enable earlier access to good medicines.
To discuss how our Catalysis and Enabling Technology groups can de-risk your current pipeline or resolve your persistent process deviations, contact our technical team today.
PS：
1.This work was published as a research article titled "Ru-Catalyzed Reductive Dehydration of Amides to Enamines: Catalyst-Loading-Controlled Selectivity" by Dr. Jacob Y. Cha*, together with the Porton team, Ziqing Zuo, William B. Reid, Todd Nelson and Thorsten Rosner; ACS Catal. 2026, 16, 4985-4992. https://doi.org/10.1021/acscatal.5c09022
END
Founded in 2005, Porton Pharma Solutions provides end-to-end CDMO services from pre-clinical to commercial stages. Our capabilities span Small Molecules, TIDES, Biologics and Conjugates (ADCs, AOCs, DACs, PDCs, RDCs, etc.), and Advanced Therapy Medicinal Products, from drug substance to drug product. As of 2025, Porton has over 4,300 employees globally, served more than 1,400 customers, and delivered over 5,000 projects. We are committed to a customer-centric approach, delivering innovative and reliable global CDMO solutions, enabling earlier access to good medicines.
This document is provided for informational purposes only and reflects relevant technologies, case studies, and industry perspectives. It should not be relied upon as a basis for drug development or other professional decisions. Porton and its affiliates make no warranties regarding the accuracy, completeness, or timeliness of the information provided and accept no liability for any use of, or reliance on, this document. Readers should make independent evaluations based on the latest scientific evidence, applicable regulations, and professional advice.
评论 (0)