The challenge posed by counterfeit pharmaceuticals

The World Health Organization (WHO) states that over 50% of pharmaceuticals ordered over the internet relate to counterfeit medicines [WHO, 2010].

The legislated way for acquiring pharmaceuticals is:

1. Obtaining a prescription from a doctor;
2. Presenting the prescription in an authorised pharmacy.

This procedure is being called increasingly into question through the internet, whereby illegal pharmaceuticals are advertised and distributed through internet pharmacies without appropriate certification. As of 2010, worldwide sales of counterfeit medicines exceeded USD 75 billion, a 90% rise since 2005 [WHO, 2010]. The internet represents for counterfeit pharmaceuticals the primary medium for communication and their distribution.

In 1992, the WHO in conjunction with the pharmaceutical industry and drug regulators developed the following definition of a counterfeit drug:

Counterfeit products may include, with the approximate percentages for each entry given in brackets:

       a) products with the correct ingredients (6%);
       b) with incorrect quantities of active ingredients (10%);
       c) without beneficial active ingredients (60%);
       d) products with harmful ingredients; (16%);
       e) or products with fake packaging (8%).

It is noteworthy that current usage employs the concept of “Substandard and Falsified (SF) Medical Products” (SSFM). With the exception of entry a) in the above list, counterfeit pharmaceuticals pose a threat, in some cases fatal, to patient health. In 3rd World countries, the health threat is greatest for counterfeits prescribed for antibiotics, malaria or AIDS. Developed countries vary in this respect in that the majority of counterfeit products target “life-style” pharmaceuticals, the most widespread of which are those used to correct erectile dysfunction.

Mention should be made of the SecurPharm initiative in Germany (SecurPharm, 2019). Pharmaceutical manufacturers, bulk pharmaceutical distributors and pharmacists are engaged in the company in order to exclude counterfeit pharmaceuticals from the legal distribution paths in Germany. Whereas this is undoubtedly a worthwhile endeavour, it does not address the extremely serious challenge posed by unregulated counterfeit pharmaceuticals that are distributed over the internet.

Given the high profitability of counterfeit pharmaceuticals, their negative global health impact and the sheer scale of their occurrence, it is of urgent importance to develop technology to distinguish them from genuine counterparts.

X-ray diffraction (XRD): a possible solution

The most comprehensive XRD study of genuine and counterfeit pharmaceuticals published to date is that of the company PanAlytical (Beckers, 2008). XRD exploits the interference effects that occur when x-rays are coherently scattered from neighbouring lattice planes of a crystalline or polycrystalline substance. These interference effects lead to so-called “Bragg peaks” in the scattered x-rays, where a peak occurs when Bragg’s equation is fulfilled:

In this equation,  is the peak order (1, 2 etc.),  is the x-ray wavelength,  is the lattice plane spacing, and  is the total scattering angle.

The study was performed on several blister packs, each containing 18 tablets, which were analysed with a commercial x-ray diffractometer. The measurement time was 10 minutes per tablet, corresponding to three hours for one blister pack. The tablets came from two main sources. Genuine tablets were provided by European pharmacies. Owing to the difficulties associated with legally procuring illegal counterfeits (see below), these were custom-manufactured to belong to the groups b) to d) listed in section 1. Although no automated data analysis was performed, visual inspection clearly revealed significant differences in the Bragg profiles of the original and counterfeit products. Beckers (2008) concluded that XRD represents an outstanding analytical technique for differentiating between authentic drugs and their counterfeit imitations.

Taylor (2009) describes the utility of the “InXitu” x-ray diffractometer for distinguishing between original and counterfeit pharmaceuticals. Although no experimental results are presented, on the basis of the work of Beckers (2008) there is no reason to question the InXitu approach. It merely represents a different technological implementation of XRD.

Several conditions must be fulfilled in order to establish XRD as a primary solution to the counterfeit pharmaceutical challenge. These include:

       a) Scan time: given the enormous quantities of pharmaceuticals that are distributed daily, a major increase in scan speed relative to conventional x-ray diffractometers of several orders of magnitude is required;
       b) Any technology for detecting counterfeit products should be able in an automated manner to compare the product that left the manufacturer with that dispensed to the end user. Owing to the highly complex pharmaceutical distribution chain, including the manufacturer, intermediary wholesalers, 3rd party re-packagers, exporters and re-importers, parallel traders and finally the dispenser (e.g. a hospital pharmacy), the final check on product authenticity should be performed as close as possible to the end user.
       c) Consider the specific case of the pharmacy of a large hospital, defined here as having ≥ 500 beds. Including inpatients, outpatients and specialist clinics, such a hospital will treat on the order of 1000 patients a day, the majority of whom will be prescribed some form of medication from the hospital pharmacy, whereby an average prescription will have several entries. The pharmacy of a large hospital will be receiving on a daily basis a huge number of packages, following that the majority of the packages will contain a multiplicity of different types of pharmaceuticals. Thus, the detection technology should be able to analyse incoming large packages having mixed pharmaceutical contents, rather than merely investigating single blister packs.
       d) These considerations illustrate the need for a scanner, based on the physical phenomenon of XRD, which however extends XRD such that it becomes fast, automated and is capable or resolving large, mixed-content packages in three spatial dimensions. These extensions are embodied in the x-ray diffraction imaging scanner described in the next section.

Characteristics of an XDI scanner

Conventional x-ray diffractometers implement the angular-dispersive XRD beam topology, in which (cf. equation 1) the angle of scatter, [\Theta], is varied whereas the wavelength, [\lambda], is held constant. In contrast to this, the x-ray diffraction imaging scanner (hereinafter referred to as XDI) implements energy-dispersive (sometimes termed “white beam”) XRD, by varying the photon energy at fixed angle of scatter. Owing to the fact that a much greater proportion of the x-ray source polychromatic beam is employed in the measurement, the energy-dispersive technique has much higher photon throughput than that of its angular-dispersive equivalent. This fact, coupled with the Multiple Inverse Fan Beam topology (MIFB), employing an x-ray multi-source, each source emitting multiple x-ray beams viewed by multiple detectors, enables an increase in photon throughput of 6 orders of magnitude relative to conventional XRD.

The XDI scanner yields a 4D data set of the scanned object comprising three spatial dimensions (voxellation) and one dimension representing the momentum transfer. The momentum transfer is defined in the following equation:

                         (2)

A comprehensive description of the XDI scanner together with representative results of its spatial and momentum resolution is given elsewhere (Harding and Isernhagen, 2018). The XDI scanner in use for security screening applications is depicted in figure 1.

Figure 1

Photograph of an XDI scanner for Checkpoint Baggage Screening applications (CBS) at a major European airport. Courtesy of MorphoDetection GmbH Hamburg, Germany.

A dramatic improvement in signal-to-noise ratio (SNR), or equivalent trade-offs in terms of enhanced spatial resolution and/or scan speed, can be achieved by fusing data from the scatter and transmission sensors of the XDI scanner, as described by Harding and Isernhagen (2019). These trade-offs are presented in equation 2:

                      (3)

In this equation, SNR is the signal-to-noise ratio; R  the 3D spatial resolution;  T  represents the time needed to scan a 3D object; and  C is a system constant, which depends on x-ray tube power, detector efficiency and the like. In words, the SNR varies as the square root of the scan time, T ; whereas the 3D spatial resolution varies as the fourth root of R.

An illustration of the 3D spatial resolution of the XDI scanner is given in the next figure.

Figure 2

Scatter strength (as the summation of a voxel’s momentum transfer profile) image of section through a water bottle having physical dimensions of 60 mm diameter and 125 mm overall length. Superimposed on the scatter strength gray scale image are white lines representing the bottle contours and a blister pack of tablets having 5 mm diameter, to give an impression of the XDI scanner spatial resolution.

Representative momentum profiles measured with the XDI scanner are presented in section 5.

Procurement of pharmaceuticals

In all, over 50 different pharmaceuticals, both genuine and counterfeited, were analysed with the XDI scanner. The two main criteria for selection of the pharmaceuticals chosen for this study were: their frequency of prescription and their profitability when counterfeited.

Genuine pharmaceuticals were procured by Pharma Consulting Services (PCS) from an accredited pharmacy in Germany. In the interests of brevity, only a summary of the pharmaceuticals, genuine and counterfeited will be given here. Notwithstanding the fact that some product names (applicable to the German market) are given below, this in no way implies their endorsement or otherwise by the authors. They are given merely for exemplary purposes.

The pharmaceuticals that were analysed by the XDI scanner will be categorized, somewhat arbitrarily, into three groups: namely those to aid a certain chosen lifestyle; those to treat recognized bodily diseases; and those to counteract certain psychological conditions. No attempt will be made to justify the inclusion of a certain pharmaceutical in one category or another, as these sometimes have fuzzy boundaries and it is beyond the scope of this article to present exact pharmacological definitions.

As mentioned previously, lifestyle pharmaceuticals offer high profit margins and are especially targeted by counterfeiters. These include medicines addressing: erectile dysfunction, such as Viagra; weight-loss, such as Xenical; anabolic steroids, such as Anabol for muscle development and performance enhancement; contraceptives such as Biviol to aid family planning; and prescription shampoos to combat hair loss and promote its growth. This last entry is interesting in that it refers to substances having liquid or more properly emulsion form at the molecular level; whereas the other lifestyle pharmaceuticals listed have a polycrystalline molecular form.

At the other end of the pharmaceutical spectrum are those medicines used to treat recognized bodily diseases. Pharmaceuticals in this category tested with the XDI scanner included: agents to reduce blood pressure, such as Lorzaar; malaria treatments, such as Cotrim; medicines to combat aids, such as Zidolam-N; and tumour growth retarders, such as Sutent.

Between the life-style pharmaceuticals and those for treating recognized bodily diseases lies the category of medicines to counteract certain psychological conditions. These include: opioid painkillers, such as Tafil; sleeping tablets, such as Tavor; and anti-depressants, such as Zoloft.

Numerous stakeholders are involved in the fight against counterfeit pharmaceuticals. In Germany, these include: federal regulatory bodies, national standardization laboratories, German pharmacist councils, pharmaceutical manufacturers, patent holders, customs authorities and police enforcement agencies. Under the auspices of PCS located in Lower Saxony, Germany, approaches were made to both national and state entities with a view to procuring representative samples of counterfeit pharmaceuticals. Whereas these approaches were received sympathetically, none of the above stakeholders felt able to provide “genuine” counterfeit pharmaceuticals for the purpose of the present study. The main reasons for this included: business sensitivities on the parts of manufacturers and patent holders; and questions regarding the responsibilities of the diverse government agencies involved for providing pharmaceuticals for testing purposes.

In order to proceed with the study, an alternative approach was made by PCS to the Institute of Pharmacy of the Friedrich-Schiller University, Jena, Germany and specifically to the department of Pharmaceutical Technology there. They were invited to compound and supply for the purpose of this study representative counterfeit pharmaceuticals in the categories (a) to (e) of section 1. Following the requirements of § 67 AMG (Arzneimittelgesetz, German Drug Law), which stipulates that every centre must give notification of its intention to test pharmaceuticals, permission to analyse with the XDI scanner in Hamburg, Germany, the tablets compounded by Jena University was sought and ultimately obtained.

XDI measurements of selected pharmaceuticals

An illustration of measurements made with the XDI scanner on authentic and counterfeit pharmaceuticals to address erectile dysfunction (belonging to the lifestyle group introduced above) is presented in figure 3.

Figure 3

Analysis with the XDI scanner of original (solid line) and counterfeited (dashed line) blister-packed pharmaceuticals addressing erectile dysfunction. The XDI profiles have been corrected for attenuation following subtraction of the background arising in an average empty blister pack.

The XDI profiles shown in this figure exhibit pronounced Bragg peaks corresponding to the predominantly crystalline structure of these pharmaceuticals. Differences in the XDI profiles, such as their peak positions, relative peak intensities etc. can be exploited as described in the next section to enable authentic and counterfeited pharmaceuticals to be distinguished.

Prescription shampoos often have colloidal form, in which solid microparticles are dispersed in an amorphous medium. Thus their XDI profiles exhibit both amorphous and crystalline features, when the dispersed particles have microcrystalline structure. An example of the XDI profile of such a mixed-phase substance is presented in figure 4. The momentum positions of the Bragg peaks can be identified with a peak-detection algorithm, such as the 2nd differential operator  which varies from positive, through zero, to negative gradient around the peak.

Following the work of Hindeleh and Johnson (1971), the amorphous background was approximated by the best-fit 3rd order polynomial, with the XDI profile intensity as the dependent and the momentum, x, as the independent variable. Then both the crystalline and amorphous phase components of the substance yield features on whose basis the substance can be classified.

Figure 4

XDI profile (solid line) of a mixed-phase substance containing both amorphous and crystalline components. The Bragg peaks of the crystalline phase can be identified with a peak detection algorithm; whereas the dashed line represents the best-3rd order polynomial fit to the amorphous background (see text).

Data analysis of XDI profiles

There are numerous types of methods for determining to which library group an unknown member belongs, such as the cross-correlation operator or Linear Discriminant Analysis (LDA). In the present case, the classification problem is simplified as it is merely necessary to determine how similar the XDI profile of a trial pharmaceutical substance is to that of the genuine substance. The methodology chosen for determining this similarity index was feature-based and followed the well-trodden path of: feature extraction; library comparison; and member classification. As noted in the previous section, the main features employed for determining the similarity or otherwise of genuine and trial crystalline pharmaceuticals were: momentum peak positions, peak shapes; and relative peak intensities.

For the mixed-phase pharmaceuticals, the list of features was supplemented by the four polynomial coefficients of the amorphous background.

Whether the pharmaceutical was single phase crystalline or mixed-phase (crystalline and amorphous), the Euclidean distance between the original and trial substance was computed. This is mathematically expressed in the following equation:

     (4)

In this equation, D is the Euclidean distance between the ith feature, i^th , F_gcharacterising the genuine pharmaceutical and the corresponding feature of the trial pharmaceutical, denoted as F_t. The distance, D, is summed over the total of N significant features and has a value that is either zero, in the case of complete identity of the genuine and trial pharmaceuticals, or positive.