Comparison of Authentic and Suspect Pharmaceuticals

Counterfeit drugs in the US drug supply are a serious public-health concern, but the scope of this problem is not known. A counterfeit drug is legally defined as any brand-name or generic product sold under a product name without proper authorization (1). By this definition, a counterfeit can be an actual product that has been relabeled or repackaged without authorization. The counterfeit products that pose the most risk to patient health are those that contain incorrect ingredients, no active ingredients, an improper dose, or harmful contaminants. Counterfeit drugs with no active ingredient or with the right active ingredient at dosages other than the labeled dosage create nonpotent, subpotent, or superpotent medications. The risks involved with taking a counterfeit drug include unexpected side effects, allergic reactions, and the worsening of the medical condition (2).

Throughout the 1990s, the US Food and Drug Administration investigated an average of five counterfeit drug cases per year. Since 2000, this number has sharply increased to more than 20 cases per year (3). In the United States, less than 1% of drugs sold at drugstores are thought to be counterfeit. Current estimates of the prevalence of counterfeit medicines range from 5–10% internationally, and as high 50% in certain countries (2, 4).

Internet pharmacies have provided a convenient way for patients to fill their prescriptions and potentially save money. Internet pharmacies also have created a new venue for the introduction of unapproved and counterfeit drugs into the US (5). The Government Accountability Office (GAO), the investigative arm of Congress, examined Internet pharmacies and found that foreign Internet pharmacies were shipping counterfeit versions of brand-name products such as Viagra (sildenafil citrate, Pfizer, New York) and Oxycontin (oxycodone hydrochloride, Purdue Pharma, Stamford, CT) to the US. GAO also found that 45 of 68 Internet pharmacies in the US, Canada, and other nations were selling prescription drugs over the counter. In addition, GAO discovered Internet pharmacies that were filling orders for painkillers with less potent but potentially addictive medications (6).

In the US, FDA and the Customs and Border Protection collaborated to perform spot examinations of suspect mail shipments to identify foreign drugs that were illegally imported. These examinations took place in July and August 2003 at postal facilities in Miami, New York, San Francisco, and Carson, California. The drugs came from countries around the world, including Canada (15.8%), India (14.3%), Thailand (13.8%), and the Philippines (8.0%) (7). Of the 1153 imported drugs examined, a total of 1019 (88%) were unapproved.

Such investigations indicate a greater prevalence of counterfeit drugs in markets other than the US, including in countries where authentic pharmaceuticals are legally authorized for sale. Regulators need simple methods to identify potentially counterfeit products. Simple visual inspection has obvious advantages and continues to function as the major initial screening technique. Progress in process analytical technologies (PAT) that have been applied to pharmaceutical manufacturing may also play a role in screening for counterfeit drugs. The authors' objective was to apply near-infrared (NIR) spectrometry to assess whether eight suspected counterfeit drug products were authentic.

Methods

Materials. The eight suspected counterfeit products listed in Table I were purchased in Hong Kong in December 2007 (8). None of the purchases required a prescription or involved a discussion with the pharmacist about medication needs. The seller of a purported version of Cialis (tadalafil, Eli Lilly, Indianapolis, IN) in a 50-mg dosage form and of Viagra in a 100-mg dosage form bargained substantially on the price, unlike the other pharmacies. Margins on medications are low in Hong Kong, and a willingness to bargain may indicate that the product is not authentic. Labeling of each of these eight test products suggested that the product was a brand product. Eli Lilly does not market Cialis in a 50-mg dosage form in the US.

Table I: Potential counterfeit products (i.e., test products).

Corresponding authentic product (except Cialis 50 mg), along with generic amlodipine besylate 10 mg tablets, were purchased from McKesson, a major distributor in the US.

Instrumentation. With some exceptions, NIR spectra were collected using the methods of Tatavarti et al. (9). The authors used an RCA XDS NIR spectrometer (Foss NIRSystems, Columbia, MD) to collect the spectra. Instrument operation and data analysis were performed using Foss's Vision 3.2 software. The data-collection method was set to full range (400–2500 nm). Samples and references were scanned in diffuse-reflectance mode an average of 32 times at 0.5-nm intervals. In addition, NIR spectra were collected using a PHAZIR-Rx instrument (Polychromix, Wilmington, MA).

NIR spectra comparison. The authors compared the NIR spectrum of each suspected counterfeit product to that of the authentic product using the ρ_m criterion expressed in the following equation (10, 11):

where R_t is the absorbance of the reference product at wavelength t,T_t is the absorbance of the test product at wavelength t, and RATIO _t is the larger of either R_t ÷ T_t or T_t ÷ R_t. Previously, ρ_m has been used to compare test and reference profiles (e.g., plasma profiles and dissolution profiles) and assess profile similarity. ρ_m is weighed by the sum of R _t and T_t. This weighting factor places more emphasis on the wavelengths with higher absorbtivities. The unweighted criterion is:

Spectra were also compared by the following Delta_a metric (10, 11):

in which δ_a is based upon the absolute difference between two spectra and considers the relative size of the profiles. For profiles that are not identical, 0 ≤ δ_a. In equation 3, δ_a is weighed by R_t + T_t. The unweighted criterion is:

Other quality tests. As discussed below, a simple visual inspection of test Cialis 50 mg resulted in the conclusion that it was counterfeit, and disintegration testing was performed using the US Pharmacopeia method. Disintegration was also conducted on the reference Cialis 20 mg for comparison purposes because authentic Cialis is not available in 50-mg strength.

In addition, when an authentic active pharmaceutical ingredient (API) was available, thin-layer chromatography (TLC) was conducted on the test and reference products. TLC was performed on amlodipine besylate, sildenafil citrate, and terbinafine hydrochloride products. Tablets were crushed and subjected to TLC in parallel with the authentic APIs. A 0.3mg/mL solution of drug in methanol was prepared for amlodipine besylate products and API and run on a TLC plate using a mixture of 2-butanone, water, and glacial acetic acid in a ratio of 50:25:25 as the mobile phase. For the sildenafil citrate products and API, a 1mg/mL solution of drug in methanol was prepared and run on a TLC plate using the same mobile phase. Terbinafine hydrochloride products and API were evaluated using the same method as that for sildenafil citrate. Plates were inspected under ultraviolet light at 254 nm.

Figure 1. Test Cialis 50 mg package. The plastic container appeared to be of low quality and failed to denote tablet quantity, manufacturer, and distributor. The packaging included a "Lilly" hologram and a misspelled "Lieel icos" marking.

Results and discussion

Visual inspection. In all cases, simple visual inspection indicated that the test product appeared authentic, compared with the reference product, except for the test product labeled Cialis 50 mg. Figure 1 shows the test Cialis 50 mg package. Figure 2 shows the test Cialis 20 mg tablet, reference Cialis 20 mg tablet, and test Cialis 50 mg tablet. The packaging for the test product labeled Cialis 50 mg, included a "Lilly" hologram and a misspelled "Lieel icos" marking. The plastic container was of low quality and failed to denote tablet quantity, the manufacturer, and the distributor.

Figure 2. Panel A shows the test Cialis 20-mg product (left) and the reference Cialis 20-mg product (right). Panel B shows the two sides of a test Cialis 50-mg product, where one side (right) exhibited poor engraving.

It should be emphasized that simple visual inspection is the dominant screening method for potentially counterfeit products. Arguably, simple visual inspection is conducted on every pharmaceutical product dispensed from a pharmacy.

Figure 3. Near-infrared spectra of Cialis 20 mg test and reference products and the test Cialis 50-mg product.

NIR spectra comparison. Figures 3–9 show the NIR spectra of test and reference products from an RCA XDS NIR spectrometer. Results from a PHAZIR-Rx instrument provided comparable results. Only mean profiles were plotted because the spectra for each product varied little. Visual inspection of spectra showed all test spectra to be similar to the correseponding reference spectra, except that of test Cialis 50 mg and test Norvasc 10 mg. Because Cialis 50 mg is not marketed in the US, it was compared with reference Cialis 20 mg. Examination of the peaks in Figure 3 showed peak shifts at 422, 473, 636, 900, 1200, 1485, 1564, 2100, and 2354 nm. Shifts in peak position indicate differences in chemical composition. Given the number of significant spectral differences, it is highly unlikely that the Cialis 50 mg is an authentic product. It should be emphasized that this conclusion stems from using NIR to screen products for authenticity without direct assistance from the manufacturer or the Hong Kong regulatory authority.

Figure 4. Near-infrared spectra of Lamisil 250 mg test and reference products.

Not surprisingly, the spectra for the authentic generic amlodipine besylate 10 mg did not match that of the reference Norvasc 10 mg (see Figure 7). Generic amlodipine besylate was manufactured by Mylan Laboratories (Canonsburg, PA). Though the generic amlodipine besylate is presumably authentic, generics typically contain different excipients than do the references, so it is not surprising that the spectra are slightly different. The vast majority of the tablet's ingredients are excipients. Furthermore, the manufacturing processes of the generic and branded product differ, thus resulting in further differences in physical attributes and increased spectral differences.

Figure 5. Near-infrared spectra of Levitra 20 mg test and reference products.

Figure 7 compares test Norvasc 10 mg with reference Norvasc 10 mg. Spectra were insufficient to determine whether the test product was authentic.

Figure 6. Near-infrared spectra of Norvasc 5 mg test and reference products.

Table II lists NIR spectra comparison results in terms of the ρ_m and δ_acriteria. Essentially, ρ_m is a ratio-based method, but δ_a is based on the absolute difference between two spectra (10, 11). Each method compares profiles at the same wavelength and can include all or selected wavelengths. If the profiles are identical, ρ_m = 0. Dissimilar profiles result in ρ_m that approaches infinity. This approach results in a ratio to assess spectra similarity, in which the larger value of either ^R t /T_t or ^Tt/R_t is considered at each wavelength. By virtue of being a ratio approach, ρ_m normalizes the differences between the profiles against the size of the profiles (i.e., it results in a value that quantifies the relative difference between the two profiles). δ_a also considers the relative size of the profiles. The authors used critical limits of ρ_m > 0.15 and δ_a > 15 to denote products that were not similar to their references because these limits represent a difference of more than 15% from the reference.

Table II: Near-infrared spectra comparisons of test product to reference product using ρm and δa.

Spectra were compared over the 400–2500-nm range and over the 800–2500-nm range. Because 400–800 nm is the visible range and color differences (e.g., between dyes) are frequently and easily imparted to products designed for different countries, the narrower 800–2500-nm range was used to minimize colorant differences.

Figure 7. Near-infrared spectra of Norvasc 10 mg test and reference products and authentic generic amlodipine besylate 10 mg.

The results of ρ_m and δ_a generally followed the results of simple visual inspection of the tablets (see Table II). Most test spectra were similar to the reference spectra (i.e., ρ_m < 0.15 and δ_a < 15), and most products appeared to be authentic. Hence, the authors concluded that test Cialis 20 mg, test Lamisil 250 mg, test Levitra 20 mg, test Norvasc 5 mg, test Viagra 50 mg, and test Viagra 100 mg were authentic. TLC results from the terbinafine hydrochloride products and API indicated that test Lamisil 250 mg contained the labeled drug. TLC results from sildenafil citrate products and API indicated that test Viagra products contained the labeled drug.

Figure 8. Near-infrared spectra of Viagra 50 mg test and reference products.

NIR comparison also agreed with simple visual inspection for test Cialis 50 mg (ρ_m = 0.303 and δ_a = 21.1 over the 400–2500-nm range). Disintegration testing of test Cialis 50 mg further supported the thesis that the product was counterfeit. Three individual tablets each required more than 15 min (15:20, 16:34, and 17:35 min) to disintegrate. Reference Cialis 20 mg disintegrated in less than 5 min (3:57, 3:57, and 4:10 min for each of three tablets).

Figure 9. Near-infrared spectra of Viagra 100 mg test and reference products.

Test Norvasc 10 mg was the exceptional product. Simple visual inspection indicated that the product was authentic, but NIR could not determine its authenticity (ρ_m = 1.07 and δ_a = 54.4 over the 400–2500-nm range). TLC results from the amlodipine besylate products and API indicated that the test Norvasc 10 mg product contained the labeled drug.

Comparison of the authentic generic amlodipine besylate 10 mg to the reference. Figure 7 plots the NIR spectra of authentic generic amlodipine besylate 10 mg. Like test Norvasc 10 mg, generic amlodipine besylate 10 mg exhibited a spectrum that was different from that of the reference (i.e., ρ_m > 0.15 and δ_a > 15). Generic amlodipine besylate 10 mg was manufactured by Mylan Laboratories. Comparing the spectral differences of the Mylan product and the test Norvasc 10 mg product with their references showed that the Mylan product differed from the reference in a pattern similar to the way the Hong Kong product differed from its reference. Each showed a baseline shift, but provided similar peak positions to references. In spite of the larger ρ_m and δ_a values for generic amlodipine besylate 10 mg and test Norvasc 10 mg, the generic product was presumed authentic, and test Norvasc 10 mg could be authentic. TLC results showed a single spot from test and reference Norvasc 10 mg products, as well as from the Mylan product, that agreed with the TLC results from amlodipine besylate API.

Conclusion

The authors concluded that test Cialis 20 mg, test Lamisil 250 mg, test Levitra 20 mg, test Norvasc 5 mg, test Viagra 50 mg, and test Viagra 100 mg were authentic and test Cialis 50 mg and test Norvasc 10 mg were potentially counterfeit. Generic amlodipine besylate 10 mg did not provide a spectrum that could be described as similar to that of the reference product, although no evidence supports the thesis that this product is counterfeit.

These results indicated that NIR is a viable approach for the quick assessment of potentially counterfeit products, particularly because NIR spectra are sensitive to both composition and physical properties. Results also indicated that initial field testing for counterfeits can be accomplished with a handheld NIR spectrophotometer. To minimize incorrect NIR-based conclusions about products, NIR spectra should be interpreted with care.

Appendix I: Further comments about ρ_m and δ_a criteria

The most similar test and reference spectra were those of Viagra 50 mg, which showed a difference of about 1%. Over the range 400–2500 nm, ρ_m = 0.0122 and δ_a = 1.21. Because of high-profile similarity above and below 800 nm, the exclusion of 400–800-nm data had essentially no effect on ρ_m or δ_a for Viagra 50 mg. Because of the high-profile similarity above 800 nm for Viagra 100 mg and the modest difference in visual tablet colors, ρ_m and δ_a for Viagra 100 mg were larger when 400–800-nm data were included. Figure 10 shows slight visual-color difference between test and referenceViagra 100 mg. The TLC results showed a single spot from test and reference Viagra products that agreed with the TLC results from the sildenafil citrate API.

Figure 10. Photograph of test (left) and reference (right) 100-mg Viagra tablets. The test and reference products were of slightly different visual color, as reflected in the 400–800-nm range in Figure 9.

Acknowledgments

The authors acknowledge the assistance of Mary Lego, education manager at Foss NIR Systems, in acquiring the spectra on a Foss NIRSystems RCA XDS NIR spectrometer.

James E. Polli* is a professor, and Stephen W. Hoag is an associate professor, both at the University of Maryland School of Pharmacy, 20 Penn St., Baltimore, MD 21201, jpolli@rx.umaryland.edu. Polli is also a member of Pharmaceutical Technology's editorial advisory board. Sharon Flank is CEO of InfraTrac.

*To whom all correspondence should be addressed.

Submitted: Oct. 1, 2008. Accepted: Dec. 15, 2008.

What would you do differently? Submit your comments about this paper in the space below.

References

1. Federal Food, Drug, and Cosmetic Act, Chapters 1–2, www.fda.gov/opacom/laws/fdcact/fdcact1.htm, accessed June 10, 2009.

2. FDA, "Counterfeit Drugs Questions and Answers" (Rockville, MD), www.fda.gov/oc/initiatives/counterfeit/qa.html, accessed June 15, 2009.

3. FDA, "New FDA Initiative to Combat Counterfeit Drugs" (Rockville, MD), www.fda.gov/oc/initiatives/counterfeit/backgrounder.html, accessed June 15, 2009.

4. World Health Organization and the International Federation of Pharmaceutical Manufacturers and Associations, "Counterfeit Drugs" (WHO, IFPMA, Geneva, 1992).

5. B.H. Murray, "Online Distribution of Counterfeit Pharmaceuticals" white paper, (Cyveillance, Arlington, VA, 2001).

6. F. Dimerik, "Bogus Pills Sold on Net, GAO Finds," Bloomberg News (Bloomberg, New York, June 18, 2004).

7. FDA, "Combating Counterfeit Drugs" (Rockville, MD, 2004), www.fda.gov/oc/initiatives/counterfeit/report02_04.html, accessed June 10, 2009.

8. S. Flank, "Anticounterfeiting and NIR: A Hong Kong Diary," Pharm. Mfg. online article (2008), www.pharmamanufacturing.com/articles/2008/068.html, accessed June 10, 2009.

9. A.S. Tatavarti et al., "Assessment of NIR Spectroscopy for Nondestructive Analysis of Physical and Chemical Attributes of Sulfamethazine Bolus Dosage Forms," AAPS PharmSciTech 6 (01), E91–E99 (2005).

10. J.E. Polli and A.M. McLean, "Novel Direct Curve Comparison Metrics for Bioequivalence," Pharm. Res. 18 (6), 734–741 (2001).

11. P.R. Seo, V.R. Shah, and J.E. Polli, "Novel Metrics to Compare Dissolution Profiles," Pharm. Dev. Technol. 7 (2), 223–231 (2002).