Challenging near infrared spectroscopy discriminating ability for counterfeit pharmaceuticals detection.

This study was initiated by the laboratories and control department of the French Health Products Safety Agency (AFSSAPS) as part of the fight against the public health problem of rising counterfeit and imitation medicines. To test the discriminating ability of Near InfraRed Spectroscopy (NIRS), worse cases scenarios were first considered for the discrimination of various pharmaceutical final products containing the same Active Pharmaceutical Ingredient (API) with different excipients, such as generics of proprietary medicinal products (PMP). Two generic databases were explored: low active strength hard capsules of Fluoxetine and high strength tablets of Ciprofloxacin. Then 4 other cases involving suspicious samples, counterfeits and imitations products were treated. In all these cases, spectral differences between samples were studied, giving access to API or excipient contents information, and eventually allowing manufacturing site identification. A chemometric background is developed to explain the optimisation methodology, consisting in the choices of appropriate pretreatments, algorithms for data exploratory analyses (unsupervised Principal Component Analysis), and data classification (supervised cluster analysis, and Soft Independent Modelling of Class Analogy). Results demonstrate the high performance of NIRS, highlighting slight differences in formulations, such as 2.5% (w/w) in API strength, 1.0% (w/w) in excipient and even coating variations (<1%, w/w) with identical contents, approaching the theoretical limits of NIRS sensitivity. All the different generic formulations were correctly discriminated and foreign PMP, constituted of formulations slightly different from the calibration ones, were also all discriminated. This publication addresses the ability of NIRS to detect counterfeits and imitations and presents the NIRS as an ideal tool to master the global threat of counterfeit drugs. Copyright 2009 Elsevier B.V. All rights reserved.