Forensic intelligence for medicine anti-counterfeiting Klara Dégardin, Yves Roggo, Pierre Margot  Forensic Science International  Volume 248, Pages 15-32 (March 2015) DOI: 10.1016/j.forsciint.2014.11.015 Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 1 Main steps of the iterative process of intelligence applied to the analysis of counterfeits (adapted from Ribaux and Margot [39]). Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 2 Main manufacturing steps of a medicine and corresponding variables and methods. (a) The steps of the production of a medicine are presented with the units of production. At each step traces are left by the counterfeiters that can be analysed [41]. (b) Each manufacturing step that is represented constitutes an analytical level for the study of counterfeits. The studied variables are displayed for each production step together with the corresponding analytical methods required. The first three steps enable to perform a chemical profiling while the last three constitute a packaging profiling. LC–MS: liquid chromatography coupled with mass spectrometry, NIR: near infrared, MIR: middle infrared, GC–MS: gas chromatography coupled with mass spectrometry. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 3 Results obtained for the LC–MS analysis of sibutramine in the counterfeit capsules. (a) Chromatogram obtained in Q3 Full scan experiment for the analysis of a counterfeit capsule coming from the seizure #6. The whole range of chromatogram can be observed with all the components of the capsule. The peak of interest, the one of sibutramine, is located at around 14min of retention time. (b) Chromatogram obtained in Precursor Ion Scan for a capsule from the seizure #6. The ion of interest is fragmented so that the isomers and/or metabolites of sibutramine can be observed. In the present case a stable fragment was measured at 125 m/z, attributed to sibutramine (peak 3). Peak 1 and peak 2 correspond to two metabolites of sibutramine, desmethylsibutramine and didesmethysibutramine. (c) Molecular formula of sibutramine, desmethysibutramine and didesmethylsibutramine. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 4 Results of the Principal Component Analysis (PCA) applied on the counterfeits and genuine samples with three Principal Components [37]. (a) Genuine products are gathered into one group. Five groups of counterfeits are also observed while other counterfeits, marked under “confused no. 1”, cannot be separated. (b) After the exclusion of the genuine and the five well-defined groups from the dataset, another PCA was computed. It resulted in the determination of four additional groups and another “confused no. 2” group. (c) The further eviction from the dataset of the four previous well-defined groups led to the observation of six supplementary groups. The three successive PCA enabled to distinguish altogether a group of genuine samples and 15 Raman classes of counterfeits. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 5 Raman spectral analysis of two profiles of seizure #27. The first profile contains, besides starch, sibutramine and taurine. The second is composed, besides of starch, of lactose, other minor saccharides and sibutramine. In the second profile, sibutramine peaks are higher than in the first, but no taurine is observable. An evolution in the chemical composition is consequently observable between the profiles of the same seizure. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 6 Results obtained with the analysis of the shell aspect of the counterfeit capsules. (a) Observation of capsule closure system dots on a counterfeit capsule (circled on the picture). The brand name and the name of the product have been masked like in the other parts of the figure. Three types of dots are observed among the counterfeits. (b) General aspect of a counterfeit capsule. Some counterfeits are compressed, whereas the general aspect of the shell is considered “normal” for most counterfeit capsules. (c) Aspect of the imprint of the brand name. While most letters are difficult to compare, the “R” is very characteristic. Some counterfeits present an “R” like in the Helvetica font type, which is also the case of genuine capsules. However some counterfeiters use a font close to the Rubrik type. (d) Three types of colours used by the counterfeiters. The colour is measured according to the colours defined in the Munsell book. Several counterfeits present a colour difficult to evaluate, between 7.5B 6/8 and 7.5B 7/8. These counterfeits were then gathered under the same “7.5B 6/8 and 7/8” type. Two other types, 5B 7/8 and 2.5B 6/6 were found among the counterfeits. The genuine capsules are attributed another type of colour. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 7 Example of the content aspect of counterfeit capsules presenting the same chemical composition and coming from three different seizures, #6 p2, #27 p2 and #22. The two capsules content on the left look similar and consequently come from the same batch. On the contrary the content of seizure #22 is made of a fine powder that suggests that the capsule comes from a different batch. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 8 Links obtained for the chemical data and box data. The 33 seizures are represented by flags and seizure information. The upper side of the figure displays chemical links while the lower side provides links between boxes. The batch numbers, box barcodes and numbers have been anonymised by arbitrary digits. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 9 Geographical repartition of the seizures. The 33 seizures studied are represented at their place of seizure. Further information like the date and type of seizure and the range of quantities seized is also available. Orders realised on the internet are symbolised by a computer and the place of seizure is in that case the country where the order was made. Other types of seizure are represented by a cross (pharmacies or clinics) and a double arrow (counterfeits seized in transit, e.g. in warehouses or at the customs). Countries where more than 1000 capsules have been seized on the whole (with all studied seizures concerned) are coloured in red. Countries where 100 to 1000 capsules were seized are in orange and for less than 100 capsules seized, the countries are in yellow. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 10 Estimation of the lifetime of the batches. For each batch that was analysed (defined on the y-axis), the date of seizure of the counterfeits belonging the batch is displayed (on the x-axis). The places of seizure are also provided. The length of the segment is therefore representative of the theoretical lifetime of the batch. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 11 Estimation of the production and distribution chain of counterfeits. Three cases are considered: capsules coming from the same batch, capsules from the same pharmaceutical unit, and capsules from different pharmaceutical units. As API is sometimes present, the step between the chemical production and the galenic production is represented as a dotted line. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions

Fig. 12 Proposed scheme for the geographical repartition of the counterfeit activities for the considered network. Regions are assigned to the chemical and galenic production, the packaging and the sale, and to the handling by wholesalers and intermediates. Forensic Science International 2015 248, 15-32DOI: (10.1016/j.forsciint.2014.11.015) Copyright © 2014 Elsevier Ireland Ltd Terms and Conditions