Scientific Direction Chemical and Physical Health Risks, Section Medicines and Health Products, Sciensano, J. Wytsmanstraat 14, 1050, Brussels, Belgium; KU Leuven, University of Leuven, Department of Pharmaceutical and Pharmacological Sciences, Pharmaceutical Analysis, Herestraat 49, O&N2, PB 923, B-3000, Leuven, Belgium.
Scientific Direction Chemical and Physical Health Risks, Section Medicines and Health Products, Sciensano, J. Wytsmanstraat 14, 1050, Brussels, Belgium; Université Libre de Bruxelles, Faculty of Pharmacy, RD3 - Pharmacognosy, Bioanalysis and Drug Research Unit, Boulevard du Triomphe, Campus Plaine, CP 205/06, 1050, Brussels, Belgium.
Talanta. 2020 Sep 1;217:121026. doi: 10.1016/j.talanta.2020.121026. Epub 2020 Apr 18.
The threats of substandard and falsified (SF) antimicrobials, posed to public health, include serious adverse drug effects, treatment failures and even development of antimicrobial resistance. Next to these issues, it has no doubt that efficient methods for on-site screening are required to avoid that SF antimicrobials reach the patient or even infiltrate the legal supply chain. This study aims to develop a fast on-site screening method for SF antimicrobials using spectroscopic techniques (mid infrared, benchtop near infrared, portable near infrared and Raman spectroscopy) combined with chemometrics. 58 real-life illegal antimicrobials (claiming 18 different antimicrobials and one beta-lactamase inhibitor) confiscated by the Belgian Federal Agency for Medicines and Health Products (FAMHP) and 14 genuine antimicrobials were analyzed and used to build and validate models. Two types of models were developed and validated using supervised chemometric tools. One was used for the identification of the active pharmaceutical ingredients (APIs) by applying partial least squares-discriminant analysis (PLS-DA) and another one was used for the detection of non-compliant (overdosed or underdosed) samples by applying PLS-DA, k-nearest neighbors (k-NN) and soft independent modelling by class analogy (SIMCA). The best model capable of identifying amoxicillin and clavulanic acid (co-amoxiclav), azithromycin, co-trimoxazole and amoxicillin was based on the mid-infrared spectra with a correct classification rate (ccr) of 100%. The optimal model capable of detecting non-compliant samples within the combined group of amoxicillin and co-amoxiclav via SIMCA showed a ccr for the test set of 88% (7/8) using mid infrared or benchtop near infrared spectroscopy. The best model for detecting non-compliant samples within the group of amoxicillin via SIMCA was obtained using mid-infrared or Raman spectra, resulting in a ccr of 80% for the test set (4/5) and a ccr for calibration of 100%. For the group of co-amoxiclav, the optimal models showed a ccr of 100% for the detection of non-compliant samples by applying mid-infrared, benchtop near infrared or portable near infrared spectroscopy. Taken together, the obtained models, hyphenating spectroscopic techniques and chemometrics, enable to easily identify suspected SF antimicrobials and to differentiate non-compliant samples from compliant ones.
劣药和假药（SF）对公共健康构成的威胁包括严重的药物不良反应、治疗失败，甚至导致抗生素耐药性的产生。除了这些问题之外，毫无疑问，需要有效的现场筛查方法来避免 SF 抗生素到达患者手中，甚至渗透到合法的供应链中。本研究旨在开发一种使用光谱技术（中红外、台式近红外、便携式近红外和拉曼光谱）结合化学计量学的 SF 抗生素快速现场筛查方法。该方法分析并使用了 58 种实际的非法抗生素（声称含有 18 种不同的抗生素和一种β-内酰胺酶抑制剂）和 14 种真正的抗生素，这些抗生素是由比利时联邦药品和保健品管理局（FAMHP）没收的，用于建立和验证模型。使用有监督的化学计量学工具开发和验证了两种类型的模型。一种用于通过应用偏最小二乘判别分析（PLS-DA）识别活性药物成分（API），另一种用于通过应用 PLS-DA、k-最近邻（k-NN）和软独立建模类比（SIMCA）检测不合规（过量或不足量）样品。能够识别阿莫西林和克拉维酸（复方阿莫西林）、阿奇霉素、复方磺胺甲噁唑和阿莫西林的最佳模型是基于中红外光谱，其正确分类率（ccr）为 100%。通过 SIMCA 检测复方阿莫西林组中不合规样品的最佳模型，使用中红外或台式近红外光谱，在测试集中的 ccr 为 88%（7/8）。通过 SIMCA 检测阿莫西林组中不合规样品的最佳模型是使用中红外或拉曼光谱获得的，在测试集中的 ccr 为 80%（4/5），校准集的 ccr 为 100%。对于复方阿莫西林组，最佳模型显示，使用中红外、台式近红外或便携式近红外光谱检测不合规样品的 ccr 为 100%。总的来说，结合光谱技术和化学计量学的获得的模型，能够轻松识别可疑的 SF 抗生素，并区分合规和不合规的样品。
