Área de Análisis de Medicamentos, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario and Instituto de Química Rosario (IQUIR, CONICET-UNR), Suipacha 531, Rosario S2002LRK, Argentina.
Área Desarrollo, Laboratorio Industrial Farmacéutico S.E. French 4950, S3006ETP, Santa Fe, Argentina.
J Pharm Biomed Anal. 2021 Feb 5;194:113786. doi: 10.1016/j.jpba.2020.113786. Epub 2020 Nov 23.
Pyrazinamide (PZA), Rifampicin (RIF), Isoniazid (ISH) and Ethambutol (ETB) form the core for the treatment of Tuberculosis, today a devastating disease in low-income populations around the world. These drugs are usually administrated by fixed-dose combination (FDC) products, to favour the patient compliance and prevent bacterial resistance. PZA exists in four enantiotropically-related polymorphs (Forms α, δ, β and γ), but only Form α is considered suitable for pharmaceutical products due to its stability and bioavailability properties. The classical approaches to address solid-state (microscopy, X-ray diffraction and calorimetry) shows limitations for quantification of polymorphs in the presence of excipients and other active components, as in the case of FDC tablets. In this work, an overall strategy was developed using near infrared spectroscopy (NIR) coupled to partial least squares regression (PLS) to quantify Form α of PZA in drug substance (raw material) and PZA/RIF/ISH-FDC tablets. For this purpose, two PLS models were constructed, one for drug substance preparing training (n = 30) and validation (n = 18) samples with a ternary composition (Form α/Form δ/Form γ), and other for FDC drug products, also including the appropriate amount of RIF, ISH and the matrix of excipients in order to simulate the environment of PZA/RIF/ISH association. The NIR-PLS models were optimized using a novel smart approach based on radial optimization (full range, 3 L V and MSC-D' and SNV-D' as pre-treatment, for raw material and FDC tablets, respectively). During the validation step, both methods showed no bias or systematic errors and yielded satisfactory recoveries (102.5 ± 3.1 % for drug substance and 98.7 ± 1.5 % for FDC tablets). When commercial drug substance was tested, NIR-PLS was able to predict the content of Form α (0.98 ± 0.01 w/w). The model for FDC tablets allowed estimating polymorphic purity in intact (0.984 ± 0.003 w/w), sectioned (0.986 ± 0.002 w/w), and powered (0.985 ± 0.004 w/w) tablets, showing the methodology could be applied to a different stage of the process (i.e premixed-powders or granulates). The suitability of the method was also verified when Form α was satisfactorily analysed in FDC fortified with Form δ and Form γ to reach 0.78, 0.88 and 0.98 w/w, Form α. This strategy results in an excellent alternative to ensure the polymorphic purity of PZA throughout the overall pharmaceutical manufacturing process.
吡嗪酰胺（PZA）、利福平（RIF）、异烟肼（ISH）和乙胺丁醇（ETB）构成了治疗结核病的核心药物，而结核病目前在全球低收入人群中是一种破坏性极大的疾病。这些药物通常通过固定剂量组合（FDC）产品进行给药，以提高患者的顺应性并防止细菌耐药性。PZA 存在四种具有各向异性相关的多晶型物（形式α、δ、β和γ），但只有形式α因其稳定性和生物利用度而被认为适合于制药产品。传统的解决固态（显微镜、X 射线衍射和差示扫描量热法）问题的方法在存在赋形剂和其他活性成分的情况下，例如在 FDC 片剂中，对多晶型物的定量存在局限性。在这项工作中，使用近红外光谱（NIR）结合偏最小二乘回归（PLS）开发了一种整体策略，以定量测定原料药和 PZA/RIF/ISH-FDC 片剂中 PZA 的形式α。为此，构建了两个 PLS 模型，一个用于药物原料的制备培训（n=30）和验证（n=18）样本，具有三元组成（形式α/形式δ/形式γ），另一个用于 FDC 药物产品，还包括适当量的 RIF、ISH 和赋形剂基质，以模拟 PZA/RIF/ISH 结合的环境。NIR-PLS 模型使用基于径向优化的新智能方法进行了优化（全范围、3 L V 和 MSC-D'和 SNV-D'作为预处理，分别用于原料药和 FDC 片剂）。在验证步骤中，两种方法均无偏差或系统误差，回收率令人满意（原料药为 102.5±3.1%，FDC 片剂为 98.7±1.5%）。当测试商业原料药时，NIR-PLS 能够预测形式α的含量（0.98±0.01 w/w）。FDC 片剂的模型允许估计完整片剂（0.984±0.003 w/w）、切片片剂（0.986±0.002 w/w）和粉末片剂（0.985±0.004 w/w）中的多晶型纯度，表明该方法可应用于工艺的不同阶段（即预混合粉末或颗粒）。当在 FDC 中加入形式δ和形式γ以达到 0.78、0.88 和 0.98 w/w 的形式α时，该方法还能够对 FDC 进行满意的形式α分析，验证了该方法的适用性。该策略为确保整个制药生产过程中 PZA 的多晶型纯度提供了一种极好的替代方法。
