×

Warning

JUser: :_load: Unable to load user with ID: 806
Sunday, 10 November 2013 22:37

Classification and quantification of API in counterfeit pharmaceuticals via Raman spectroscopy

Written by
Rate this item
(1 Vote)
Si v Hi plot of samples after moving average smoothing, SNV and Savitzky-Golay 1st derivative pre-processing against the PCA-C05 model at 0.5, 5 and 25% p-values for Si. Samples classify as 5 mg Cialis if they sit within the Si and Hi bounds. Si v Hi plot of samples after moving average smoothing, SNV and Savitzky-Golay 1st derivative pre-processing against the PCA-C05 model at 0.5, 5 and 25% p-values for Si. Samples classify as 5 mg Cialis if they sit within the Si and Hi bounds.

Background

Counterfeit medicines have become an increasing issue worldwide, affecting both developing and developed countries. The presence of counterfeit medicines have a wide range of impacts including health, economic and social effects.[1-4] A major source of counterfeit medicines is sales via the Internet where it has been estimated that medicines purchased from Internet sites that conceal their actual physical address are counterfeit in over 50% of cases.[5]

 

This raises concern because increasing numbers of consumers are using Internet sites to buy pharmaceuticals. Expensive, patented and unsubsidised `lifestyle' medicines, such as those used for erectile dysfunction and weight loss are often significantly cheaper when purchased online.[6]

Due to the serious health risks of counterfeit and substandard medicines, efficient detection and seizure of these products are a public health issue. Analysis techniques that are non-destructive, simple to use, cost-effective and reliable, potentially in a non-laboratory based environment, are required. The risk of the product depends on both the identity and amount of ingredients, including the active pharmaceutical ingredient (API), and therefore, qualitative and quantitative analysis is desirable.

Raman spectroscopy in cohort with multivariate data analysis techniques is a valid candidate for counterfeit detection. Raman requires minimal sample preparation is a non-destructive technique which yields information on both the active pharmaceutical ingredient and excipients.

In this study, we simultaneously performed a qualitative and quantitative analysis of counterfeit and unregistered (and potentially substandard) medicines using Raman spectroscopy. The technique was used to a) detect counterfeit medicines, b) identify the API present, and c) quantify the API present. The products analysed were those purporting to contain the API, tadalafil (innovator product marketed as Cialis®), and had been ordered on-line and intercepted at the New Zealand border. To optimise the analyses, we have combined Raman spectroscopy with a systematic study of different preprocessing techniques and classification systems for the qualitative and quantitative analysis of the intercepted products.

The results of this study can be applied to better realise the potential of laboratory based and in-filed hand-held Raman devices alike for rapid analysis of medicines in a range of settings.

Classification of genuine Cialis

Classification of genuine Cialis® from generic and counterfeit samples was obtained using soft independent modelling of class analogy (SIMCA). The best preprocessing method for classification depended on the sample classified. The classification of 10 and 20 mg Cialis® worked best with multiplicative scatter correction (MSC) as the preprocessing method. The classification of 5 mg Cialis® worked best with linear baseline correction (LBC) in combination with standard normal variate (SNV) and Savitzky-Golay 1st or 2nd derivatives. The Si versus Hi plot for 5 mg Cialis detection is shown in Figure 1 where Classification of genuine from counterfeit and generic samples was highly dependent on the preprocessing technique used. 

The classification techniques partial least squares discriminant analysis (PLS-DA), principal component regression (PCR) and support vector machines (SVM) were not able to classify genuine from generic and counterfeit samples. This is thought to be associated with the generic and genuine products being very close in composition, yet having very different classifications, and the generic and counterfeit samples having a very different composition but the same classification. SIMCA uses a more narrow classification for belonging, and if a sample isn’t close enough to the model centre then it fails classification.

Classification and Quantification of tadalafil and sildenafil tablet concentrations

Classification to determine the API in a given samples was effective with PLS-DA, PCR and SVM classification methodologies. The preprocessing methodologies that tended to work the best across these classification methods was LBC in combination with either SNV or MSC. SIMCA failed to classify based on API.

To determine the danger of a medicine the approximate concentration of API can be very useful. If a tablet has high levels of API such as tadalafil the potential harm to the patient is high especially if not being taken under the guidance of a medical professional. In contrast, if there are very low or sub-therapeutic levels of the API tadalafil then the potential harm to the consumer is much lower (the potential for harm however, is high, for other medicines if non-treatment is in itself dangerous).

PLS-R was then used to predict API concentration using average HPLC results of the counterfeit and generic samples to determine the API concentration. As the tablet to tablet variation between some counterfeit samples was very high (some samples differed by 40 mg/tablet), the errors in the model were enhanced. The surface spectra model over the spectral region 1750-1440 and 950-670 cm-1 with LBC and SNV preprocessing performed adequately with a maximum error of ± 22 mg/g (which is from the highly inhomogenous sample set #23)

Conclusions

Raman spectra of 23 batches of genuine, generic and counterfeit Cialis tablets were taken and analysed by a range of different preprocessing parameters and classification techniques. SIMCA classification in combination with three different preprocessing combinations was found to be most effective in classifying samples into genuine or generic/counterfeit classification groups. When distinguishing between samples API the classification methods PLS, PCR and SVM were effective in combination with baseline correction and SNV preprocessing.

The API concentration of counterfeit medicines can be determined using Raman spectroscopy, this can be useful in determining the potential risk a counterfeit drug carries with respect to the API levels.

The work was published in the Journal of Raman Spectroscopy (2013) 44, 1172 – 1180 and was highlighted in Spectroscopy Solutions (http://www.spectroscopy-solutions.org/Information/Archive?pgid=243&artid=2349)

References

1. Issack, M.I., Substandard drugs. Lancet, 2001. 358(9291): p. 1463-1463.

2. Alubo, S.O., Death for Sale - a Study of Drug Poisoning and Deaths in Nigeria. Social Science & Medicine, 1994. 38(1): p. 97-103.

3. Kao, S.L., et al., An Unusual Outbreak of Hypoglycemia. New England Journal of Medicine, 2009. 360(7): p. 734-736.

4. Newton, P.N., et al., Counterfeit anti-infective drugs. Lancet Infectious Diseases, 2006. 6(9): p. 602-613.

5. IMPACT. Counterfeit medicines: an update on estimates. 2006 [cited 2012 March]; Available from: http://www.who.int/medicines/services/counterfeit/impact/TheNewEstimatesCounterfeit.pdf.

6. Baert, B. and B. De Spiegeleer, Quality analytics of internet pharmaceuticals. Analytical and Bioanalytical Chemistry, 2010. 398(1): p. 125-136.

PSSRC Facilities

The Otago group have a number of facilities around spectroscopy with a Raman microscope, FT Raman facility and low frequency facility. The low frequency facility has been used to look at recrystallization as a complementary method to THz. We have also just got a handheld NIR spectrometer and are looking to use that on some of the counterfeit materials. 

Read 32062 times