Conjugation of protein therapeutics with polymers like polyethylene glycol (PEG) has been shown to increase their therapeutic efficiency. However, manufacturing of PEGylated drugs requires an additional, carefully controlled reaction step after purifying the protein, followed by further purification of over- and under-PEGylated variants. In this work, we have used a combined spectroscopic and statistical approach for monitoring and control of the PEGylation reaction for G-CSF using near infrared spectroscopy (NIRS). An online NIRS probe deployed in the reaction vessel has been used to track conversion of G-CSF into monoPEGylated and multiPEGylated forms using calibrated partial least squares regression models on the NIRS spectra which are collected in real time every 3 seconds. A pH probe integrated with a peristaltic pump facilitates automated quenching of the reaction at the targeted time. The NIRS spectra have also been used to build a batch evolution model for the reaction from end-to-end, including the addition of the reactants to the reaction vessel, the progress of the reaction for 70 minutes, and the final quenching with Tris base. Online spectra are compared against the statistical process control charts of the batch evolution model in real time to detect deviations as soon as they occur. The system was demonstrated for four common deviations in the PEGylation process, namely: delayed quenching time, wrong concentration of reducing agent added, wrong PEG to G-CSF ratio, and wrong sequence of addition of reactants. The system was able to identify all four deviations in real time and alert the operator to take control actions. The PAT approach suggested here embraces the quality by design framework and can be generalized for manufacturing scale monitoring and control of different biotechnology reactions with spectroscopic signatures.

蛋白质治疗剂与聚乙二醇 (PEG) 等聚合物的结合已被证明可以提高其治疗效率。然而，PEG化药物的生产需要在纯化蛋白质后进行额外的、仔细控制的反应步骤，然后进一步纯化过聚乙二醇化和聚乙二醇化不足的变体。在这项工作中，我们使用组合光谱和统计方法来监测和控制 G-CSF 的聚乙二醇化反应，使用近红外光谱 (NIRS)。部署在反应容器中的在线 NIRS 探针已被用于跟踪 G-CSF 向单聚乙二醇化和多聚乙二醇化形式的转化，使用校准的偏最小二乘回归模型对每 3 秒实时收集一次的 NIRS 光谱。与蠕动泵集成的 pH 探针有助于在目标时间自动淬灭反应。NIRS 光谱也被用于构建端到端反应的批次演化模型，包括将反应物添加到反应容器中，反应进行 70 分钟，最后用 Tris 碱淬灭. 在线光谱与批次演化模型的统计过程控制图进行实时比较，以在偏差发生时立即检测。该系统针对 PEG 化过程中的四个常见偏差进行了演示，即：延迟淬灭时间、添加的还原剂浓度错误、PEG 与 G-CSF 的比例错误以及反应物的添加顺序错误。该系统能够实时识别所有四个偏差，并提醒操作员采取控制措施。这里建议的 PAT 方法包含质量源于设计框架，并且可以推广用于具有光谱特征的不同生物技术反应的制造规模监测和控制。