EP1307730A1

EP1307730A1 - Quality control and standardisation of tobacco by means of nmr and pattern recognition

Info

Publication number: EP1307730A1
Application number: EP01954180A
Authority: EP
Inventors: Peter John Hylands
Original assignee: High Value Horticulture Ltd
Current assignee: High Value Horticulture Ltd
Priority date: 2000-08-04
Filing date: 2001-08-06
Publication date: 2003-05-07
Also published as: TW544286B; WO2002012872A1; AU2001276526A1; GB0019264D0

Abstract

A process for establishing a standard specification for a tobacco plant material, comprises: (i) preparing a test solution or test extract of a sample of the tobacco plant material which is known to possess the or each property required for the standard; (ii) submitting the said solution or extract to analysis by a combination of NMR spectroscopy and a computer-based pattern recognition technique; (iii) obtaining results from the analytical methods used in step (ii); and (iv) establishing a standard specification for the said tobacco plant material on the basis of the results obtained in step (iii). Candidate samples of the tobacco plant material may subsequently be tested for compliance with the standard. They can be accepted or rejected depending on whether they five analytical results which fall within or outside either part or all of the specification established in step (iv).

Description

QUALITY CONTROL AND STANDARDISATION OF TOBACCO BY MEANS OF NMR AND PATTERN RECOGNITION

The present invention relates to the use of NMR spectroscopy in combination with computer-based statistical procedures in the standardisation and quality control of tobacco.

The taking of tobacco, by either inhalation or ingestion, is a recreational activity enjoyed by a significant sector of most societies around the world. Typically tobacco in leaf form is smoked, for instance in cigarettes, cigars or pipes, but it may also be chewed. Alternatively, but less commonly, the leaves can be pulverised for talcing in powder form as snuff.

Tobacco is grown as a cash crop in many parts of the world including the Americas, southern Africa, Russia and India. The tobacco leaves are harvested and cured and are then typically bulked, graded and baled before being left to mature for one or two years prior to commercial processing. "Tobacco" refers to plants of the genus Nicotiana. N. tabacum is now the principal species used for commercial tobacco although N.rustica was the species originally introduced into Europe in the 17^th century. N.rustica contains a high level of nicotine and is now only cultivated on a commercial scale in Russia and northern India. Species of Nicotiana are in any event highly polymorphic and a given species is best regarded as an assembly of cultivars. Cultivars are divided into classes according to the method used for their curing which, as described below, may be (i) flue curing, (ii) fire-curing, (iii) air-curing, or (iv) sun-curing.

The principal example of a flue-cured cultivar is Orinoco from which the following cultivars have arisen: White stem Orinoco; Virginia Bright; 400; Oxford;

Dixie Bright; Vamorr; Vesta; Coker; White Gold; and Kutsaga 51. The principal fire-cured type cultivar is Pry or, which originated from the flue-cured Orinico. Another fire-cured cultivar is Heavy Western. In the air-cured category Burley is the principal type cultivar. Others include Southern Maryland and Maryland Mammoth. Flue-curing is typically carried out in brick barns in which heat from wood, oil, gas or other sources is supplied in enclosed flues. It is essential for the temperature and the humidity in the barn to be controlled. Fire-curing is usually carried out in log or grass barns. The tobacco leaf is hung for 4 to 7 days to turn yellow and small open fires are then made in pits in the floor to provide smoke and to produce the creosotic and distinctive aroma. Air-curing is a natural process, being carried out under normal atmospheric conditions in a wood or grass barn. The tobacco leaf should yellow before it dries out after which the rate of drying is gradually increased by increasing the ventilation in the barn. Finally, sun-curing is a process of exposing tobacco leaves to the sun. It is useful to air-cure leaves for 2 to 3 days and ferment them it in heaps for 24 to 36 hours before exposing them to the sun. Sun-curing is used in the production of Turkish tobacco.

After curing by one of the above methods, the tobacco leaves are taken from the barn and bulked. This is typically carried out on platforms with weights placed on the bulks. The bulks should be large enough to allow some fermentation of the tobacco. Bulking needs to take place for at least a month before the tobacco leaves are graded (according to size, colour and texture) for marketing. The grade of the tobacco leaf depends partly on the position of the leaf on the plant. The bottom leaves are called lugs, the lower middle leaves cutters, and the upper, middle and top leaves are called leaf tobacco.

The identity of the tobacco cultivar, the grade of the leaf and the method of curing all contribute to the characteristic properties of a given tobacco. Those properties will determine the appropriate end use for the tobacco. The main commercial products in which tobacco is used are cigarettes, cigars and pipe tobacco.

Cigarettes are produced mainly from flue-cured tobacco, optionally with a proportion of light air-cured or Turkish tobacco added. During manufacture various additives may be used as flavouring and/or conditioning agents. Typically these are formed into a "sauce" which is applied to the tobacco leaves either by immersion or by spraying. The types of additive used include glycerine, licorice, sugar, molasses, menthol and tonka bean (Dipterix odorata).

Cigars are typically made from air-cured tobacco leaf and special cultivars are used. Pipe tobacco is usually made from blended flue and air-cured leaf. Snuff is also typically prepared from dark air and/or fire-cured tobacco. As is apparent from the foregoing, many factors contribute to the overall quality and taste of a finished tobacco product such as cigarettes. These include the particular cultivar(s) used, the growing and harvesting conditions, the curing and processing techniques employed and the additives (if any). Manufacturers of tobacco products set great store by achieving a consistent taste and quality for the goods which they market under particular brand names. Indeed, most tobacco users favour one or two commercial brands and will choose those to the exclusion of most others on the grounds that the strength, flavour and other characteristics of the product are best suited to their own individual tastes. For this reason any given brand of cigarette, cigar or pipe tobacco tends to have a loyal following of consumers.

The nature of the market for tobacco products is therefore such that counterfeit branded goods present a particular threat to the authentic manufacturers. Whilst undercutting price, counterfeit manufacturers also undermine the reliability and consistency of the genuine branded tobacco products because the quality and properties of the cigarettes, cigars and pipe tobacco which they sell in lookalike packaging are significantly different from those of the genuine article.

A related problem is that of adulteration, whereby the tobacco of the required grade and quality for a given branded product is supplemented by unauthorised manufacturers with cheaper tobacco of lower quality. As discussed above, this practice serves to undermine the reliability and consistency of the branded product as perceived by the consumer as well as undercutting the authentic manufacturer.

At present, tobacco can be analysed chemically for three of its constituents. These are nicotine, chloride (ions) and total sugars. The three analyses are performed by HPLC and GLC. In the commercial blending of tobaccos an additional organoleptic evaluation is frequently carried out, the purpose of which is to distinguish between different batches of a given blend of tobacco or between different blends. However , this relies on an individual's qualitative discernment of subtle differences in aroma and is thus purely subjective. Neither the chemical analysis nor the organoleptic evaluation provides a reliable and consistent means of distinguishing between types and blends of tobacco and tobacco products. There is therefore no technique available at present which permits authentic tobacco products to be reliably and consistently differentiated from ostensibly identical products sold by counterfeit producers under the same brand name. There is no convenient method of verifying the authenticity of a branded tobacco product or of confirming that a suspected counterfeit tobacco product is indeed a fake.

It is also virtually impossible at present to provide any assurance that samples of a given tobacco material obtained from disparate sources possess a uniform quality, or that different batches of a given blend of tobacco are consistent. Similarly there is no means of discriminating reliably between different blends of tobacco or between different batches of a given tobacco product.

There is thus a need in the tobacco industry for a convenient means of auditing the quality of tobacco plant material at all stages of its post-harvesting treatment and commercial production; for distinguishing between different batches and different blends of tobacco plant material; and for monitoring the origin, cultivar type and other properties of tobacco plant material.

The present invention addresses this need by providing a method for the standardisation and quality control of tobacco plant material. The approach takes account of the totality of the components of the tobacco plant without demanding any inquiry into the intrinsic nature of either the components themselves or the tobacco plant's biochemistry. As used herein the term "tobacco plant material" encompasses tobacco plants as such, processed tobacco leaves and finished tobacco products such as cigarettes, cigars and pipe tobacco.

Thus the process of the present invention provides a means of defining a standard for a given tobacco plant material on the basis of a known sample of the material which possesses the particular property desired for the standard. A specification for the standard is established by submitting the known sample to a combination of NMR spectroscopy and a computer-based pattern recognition technique and defining the results thus obtained as the standard specification. Subsequent "candidate" samples of the said tobacco plant material can then be tested for compliance with the standard. They can be accepted or rejected depending on whether they give analytical results which fall within or outside either part or all of the established specification,

The present invention accordingly provides a process (A) for establishing a standard specification for a tobacco plant material, the process comprising: (i) preparing a test solution or test extract of a sample of the medicinal tobacco plant material which is known to possess the or each property desired for the standard; (ii) submitting the said solution or extract to analysis by a combination of NMR spectroscopy and a computer-based pattern recognition technique; (iii) obtaining results from the analysis of step (ii); and

(iv) defining a standard specification for the said plant material on the basis of the results obtained in step (iii).

The standard specification resulting from step (iv) is thus based on the results of NMR spectroscopy and computer-based pattern recognition. The invention further provides a process (B) for providing a sample of a tobacco plant material, which sample complies with a previously defined standard specification for that plant material, the process comprising:

(i') preparing a test solution or test extract of a candidate sample of the tobacco plant material; (if) submitting the said solution or extract to analysis by a combination of NMR spectroscopy and a computer-based pattern recognition technique; (iii') obtaining results from the analysis of step (ii'); and (iv') selecting the candidate sample if the results obtained in step (iii') comply with the standard specification for the said material established in step (iv) of the process above.

Likewise the invention provides a process (C) for identifying and rejecting a sample of a tobacco material which fails to comply with a previously defined standard specification for that material, the process comprising.

(i⁵) preparing a test solution or test extract of a candidate sample of the tobacco plant material;

(if) submitting the said solution or extract to analysis by a combination of NMR spectroscopy and a computer-based pattern recognition technique; (iii') obtaining results from the analysis of step (ii'); and (iv') rejecting the candidate, sample as sub-standard if the results obtained in step (iii') do not comply with the standard specification for the said material established in step (iv) of the process above. Processes (B) and (C) are conveniently carried out on a high-throughput batch scale. Candidate samples are taken from batches of the same tobacco plant material and submitted to steps (i') to (iv'). Process (C) is especially useful for detecting a sample of a tobacco product which is suspected of being a counterfeit or of being adulterated with non-authentic tobacco. The product may be, for instance, a branded cigarette, cigar or pipe tobacco.

The "property desired for the standard " in the context of above process (A) of the present invention may be any property or quality possessed by, or attributed to, a tobacco plant material. Examples of this include a recognised commercial quality, for instance that of a branded tobacco product; a particular tobacco cultivar or variety; an authenticated origin (in terms of either tobacco growing location or commercial batch); and a particular pathological state. The pathological state in question may be a given level of maturity, dictated for instance by the time of harvesting, the position of the leaf on the tobacco plant or an established resistance to a parasite, herbicide, insecticide or other agent with potential for damage to the tobacco plant in question.

In a preferred aspect of the invention the sample of tobacco plant material which is "known to possess the or each property desired for the standard" is a sample of authenticated and audited tobacco plant or finished tobacco product of which the provenance is known. For instance, it may be a sample of a specified tobacco cultivar, a sample of tobacco of a specific leaf grade, a sample of tobacco grown in a particular geographic location or a sample of tobacco which has been cured by a particular technique. It may alternatively be a sample of an authentic branded finished tobacco product. A standard specification is established by submitting that sample to NMR spectroscopy/pattern recognition as described above. Subsequent samples of the same tobacco plant material, the origin or quality of which is not known or is in doubt, can then be tested for compliance with the standard specification thus established for the authenticated and audited material.

The process of the invention represents a significant departure from the conventional technique of analysing tobacco for specific isolated components such as nicotine or sugar. Thus, in contrast, the present process yields data which take account of more than one compound, or class of compound, present in the tobacco plant material.

The solvent used for the test solution or test extract is typically one which can dissolve more than one compound or class of compound present in the tobacco plant material. Preferably it is one which can solubilise members of the maximum number of different classes of compound present in the tobacco plant material . The classes of compound mentioned in this context include phytochemicals and other categories of compound and may be, for instance, classes based on structure or on function. In one embodiment the test solution or test extract therefore contains more than one compound present in the tobacco plant material. In another embodiment the test solution or test extract contains members of more than one class of compound present in the tobacco plant material, for instance members of more than one structural class or functional class. The test solution or test extract will preferably contain representatives of the maximum number of classes of compound present in the tobacco plant material which respond to the NMR technique being used. It is thus possible for the NMR data obtained in the process of the invention to reflect the maximum number, preferably the totality, of compounds or classes of compound present in the tobacco plant material which are responsive to the NMR technique being used.

Typically the test extract is a total extract. A preferred solvent for the test solution or test extract is methanol. As will be apparent from the description above, it is preferable to avoid the use of an extract of a single isolated compound or class of compound in the process of the invention. However, such use is nonetheless embraced within the scope of the invention . The process is preferably, but not necessarily, carried out without prior selection, optimisation or isolation of a specific component of the tobacco plant material.

Nuclear magnetic resonance spectroscopy (NMR) is known by itself as an analytical tool in the investigation of plant materials. One example of its application is in the verification of the authenticity of drinks derived from fruit. In one approach hydrogen-2 NMR spectroscopy has been employed with the technique of site- specific natural isotope fractionation (SNIF-NMR) as a means of establishing the authenticity of fruit juices. For instance, in JAOAC Int. 1996 Jul-Aug, 79(4): 917- 928 Martin et al describe the use of hydrogen-2 NMR spectroscopy (SNIF -NMR method) to detect fruit juices which have been adulterated with added beet sugar. The technique relies on the fact that, when a fruit juice or fruit concentrate is fermented, the proportion of the resulting ethanol molecules which are mono- deuterated on the methyl site decreases with the addition of beet sugar. Thus any fruit juice sample to which beet sugar has been added will have a significantly lower (D/H) isotope ratio than a corresponding authentic sample. This technique has also been applied to the detection of wine chaptalisation using hydrogen-2 NMR spectroscopy, as reported for example in J.Chim. Phys.-Chim Biol. 1983, vol 80, pp 293 - 297 by Martin et al.

Hydrogen-1 NMR spectroscopy cannot itself conveniently be applied to plant materials because it generates spectra that are too complicated to be analysed visually. A solution to this problem, reported for instance by Kowalski and Bender in J. Am. Chem. Soc. 1972, 94 , 5632 - 5639, is to analyse the data by appropriate multivariate statistical analysis, for example principal component analysis (PCA). This is a technique of pattern recognition where the dimensionality of the data is reduced by combining correlated variables (peaks in the spectrum) to form a new smaller set of independent orthogonal variables called principal components (PCs).

These PCs are ordered according to their ability to explain the variance contained in the original data. A projection of the samples into a space spanned by the first PCs provides an insight into the similarity or dissimilarity of the samples with respect to their biochemical composition. Unknown or test samples can also be projected onto this space and can thus often visually be compared with the reference samples (Vogels et al, J. Agric. Food Chem. 44, 175 - 180, 1996). The combination of hydrogen- 1 NMR spectroscopy with pattern recognition techniques has been applied as a screening tool in determining the authenticity of orange juice (Vogels et al, 1996 loc. cit.). The adulteration of suspect samples could be detected by this means. The identity of the responsible contaminants was then determined by correlation of the PCA results with particular resonances present in the original NMR spectrum.

A combination of hydrogen- 1 NMR spectroscopy and carbon- 13 NMR spectroscopy with PCA has also been used to differentiate wines on the basis of their origin (Vogels et al, Chemometrics and Intelligent Laboratory Systems: Laboratory Information Management, 21 (1993) 249 - 258). Discriminant plots of samples originating from different wine-producing regions in Germany showed clustering of the samples by origin in the discriminant space after a supervised method of statistical analysis. Subsequently, reconstructed spectra were prepared from the PCA data to reveal the NMR spectroscopic peaks of the particular wine constituents (for instance monosaccharides such as glucose, mannose, rhamnose and galactose) responsible for the differentiation. Similar studies are reported elsewhere, for instance by Vogels et al in Trends in Flavour Research, Maarse & Van de Heij (Eds), Elsevier, Amsterdam (1994) pp 99 - 106.

Another application of hydrogen- 1 NMR spectroscopy and principal component analysis is reported by Trevisan et al in. Chapter 8 of "Characterisation of cell suspension cultures of hop, Humulus lupulus L.", a thesis presented to the University of Leiden, the Netherlands, pages 95 - 122, published in 1997. The authors carried out hydrogen-1 NMR spectroscopy and PCA on treated cell extracts with the aim of identifying specific metabolites accumulated by the cells following treatment. They were therefore interested, in following specific peaks in the NMR spectrum which were known to be due to individual cell components.

In contrast to these reported methods the NMR spectroscopic and pattern recognition procedure employed in the process of the present invention requires neither an investigation into the biochemistry of the tobacco plant material being analysed nor a subsequent correlation of the pattern recognition results with particular NMR spectroscopic resonances attributed to specific component(s) of the tobacco plant material. Instead it relies upon the information presented by the inherent pattern of clusters derived from NMR data, those data in turn reflecting the totality of the compounds in the tobacco plant material which respond to the NMR spectroscopic technique being used. The NMR spectroscopy combined with computer-based pattern recognition

(hereinafter termed NMR spectroscopy/pattern recognition) employed in the process of the invention typically comprises:

(a) submitting the test solution or test extract to NMR spectroscopy and recording one or more NMR spectra; and (b) submitting the data obtained from the or each NMR spectrum to a multivariate analysis to generate one or more points on a score plot. A sphere of acceptability is typically defined around the point or points on the score plot generated in step (ii) above when the NMR spectroscopy/pattem recognition analysis is being used to establish a standard specification for a tobacco plant material. That sphere then constitutes part of the specification. Candidate samples of the same material are subsequently accepted or rejected depending whether, when submitted to the NMR spectroscopy/pattern recognition analysis defined above, they give points in step (ii) which fall within or outside the sphere.

The NMR spectroscopy/pattern recognition can be used by itself as a way of standardising samples of tobacco plant materials. Accordingly, in one aspect the present invention provides a process for providing a sample of a tobacco plant material which complies with a previously established standard specification for that material, the process comprising:

(i") preparing a test solution or test extract of a candidate sample of the said tobacco plant material;

(ii") submitting the test solution or test extract to NMR spectroscopy and recording one or more NMR spectra; (iii") submitting the data obtained from the or each said NMR spectrum to a multivariate analysis to generate one or more points on a score plot; and (iv") selecting the candidate sample as a sample which complies with the said standard specification only if the points generated on the score plot in step (iii) fall within a sphere of acceptability as defined in the standard specification.

The standard specification in this aspect of the invention may be provided by a process which comprises: (i " ' ) preparing a test solution or test extract of a sample of the said tobacco plant material which is known to possess the or each property desired for the standard; (ii" ') submitting the test solution or test extract to NMR spectroscopy and recording one or more spectra; (iii'") submitting the data obtained from the or each said NMR spectrum to a computer-based multi-variate analysis to generate one or more points on a score plot; and (iv'") defining a sphere of acceptability around the points generated in step (iii'") as the, or as part of the, standard specification for the said tobacco plant ^* material.

In general multivariate analysis theory, the total data is the product of the scores multiplied by the loadings. The loadings plot can be used to define the contribution of each of the variables (spectral descriptors). A "score plot" is a graphic representation in which samples are projected into the space spanned by two or more principal component axes. Principal component analysis (PCA) is a particular method used to analyse data included in a multivariate analysis. In PCA the position of the samples can be plotted in a score plot in two dimensions where similar samples will tend to form clusters while dissimilar samples will tend to spread over large distances (Kowalski & Bender, 1972, loc. cit. and Trevisan, 1997, loc. cit.).

The context in which the points are generated on the score plot in the process of the present invention must be the same when establishing the standard specification as when analysing candidate samples for compliance with the standard. The components of the methodology used to establish the positioning of the point or points on the score plot for the known sample used to define the standard must be present when the NMR spectroscopic data from the candidate test samples are processed. In practice the data derived from the NMR spectrum of the sample used as the standard are subjected to appropriate manipulation by multivariate statistical methods, for example principal component analysis or canonical variation, together with those of the standard. The sphere of acceptability is defined by limits in the score plot which have been established on the basis of the position in the score plot of points derived from one or more extracts of the known sample. In a preferred aspect of the invention the multivariate analysis is performed using an unsupervised methodology.

The NMR spectroscopic technique used in the invention may involve carrying out hydrogen- 1 NMR spectroscopy at high fields in combination with multivariate analysis. In this particular aspect the NMR spectra are typically measured at 400 to 700 MHz. The data derived from them are then analysed by multivariate analysis software, for instance the commercially available "Pirouette" package. Examples of the high resolution hydrogen- 1 NMR spectroscopic and pattern recognition analysis are discussed by M. L. Anthony et al in Biomarkers 1996, 1, 35- 43 and Molecular Pharmacology 46, 199- 211, 1994, and by J.O.T. Gibb et al in Comp. Biochem. Physiol. vol. 118B No. 3, pp 587- 598, 1997.

As an alternative to 1 -dimensional high field hydrogen- 1 NMR spectroscopy, 1 -dimensional NMR spectroscopy using other NMR-active nuclei such as carbon- 13 or hydrogen-2 may be used in the present invention. It is also possible to use a range of 2-dimensional pulse sequence spectroscopic investigations with hydrogen- 1 or other NMR-active nuclei such as those mentioned above. The same principles apply in each case, though, and the results are analysed by appropriate multivariate analysis.

The NMR spectra may be normalised or non-normalised before the computer- based pattern recognition is carried out. Normalisation has the effect of removing peak intensity, which is a purely quantitative parameter of the spectra, as a discriminating factor. Normalisation is therefore typically carried out when the main objective of the procedure is to highlight qualitative differences between spectra obtained from different samples. However, in some cases peak intensity may be required as a discriminating factor when absolute quantitative values, for instance potency, are required. In such situations the spectra are non-normalised. .

An important advantage of NMR spectroscopy/pattern recognition is that it is not limited by a selective delivery or detection system. Spectra can be recorded without prior purification of the test solution or test extract, thus allowing all components of the tobacco material sample which possess a proton to contribute to the overall NMR spectrum. Analysis of the spectrum by the multivariate analytical techniques discussed above reveals potential valuable discriminating features of the spectra which can be used with a high degree of precision for the description of the complex mixtures of components contained in tobacco plant materials.

It is nonetheless possible in certain cases for the differentiation of samples of tobacco plant material on the score plot to be poor, with points deriving from identical samples of a given tobacco plant material being spread widely rather than forming a cluster. This loss of similarity arises when there is variation in the NMR spectroscopic shift values of individual components of the plant material. Such variation may be caused, for instance, by the presence of an overwhelmingly high concentration of one particular compound in the tobacco or by the modifying effect of pH or metal ions which cause shift values to change. If this problem occurs the test solutions or test extracts of the tobacco plant material which are submitted to NMR spectroscopy may be pre-treated to remove the source of error and achieve better clustering in the score plot. In one aspect the process of the invention therefore includes the additional initial step of purifying the test solution or test extract of the candidate sample of tobacco plant material prior to submitting it to NMR spectroscopy. Tea provides a convenient illustration of this principle. Signals due to caffeine dominate the hydrogen- 1 NMR spectrum of tea and so when NMR spectroscopic results of tea are processed by multivariate analysis the points on the score plot for samples of the same tea do not form clusters. If caffeine is removed from tea prior to carrying out NMR spectroscopy, for instance by reverse phase chromatography, proper clustering of the samples occurs and the similarities between like samples on the score plot become clear. This is illustrated in Example 2 and accompanying Figures 4 A and 4B. Figure 4A is a score plot for untreated tea samples where clustering is indistinct. Figure 4B is a score plot for pre-treated tea where, in contrast, there is clear clustering which allows positive discrimination. The NMR spectroscopy/patte recognition analysis is highly sensitive and has the capacity to differentiate samples of tobacco plant material which appear to be identical when analysed by other methodologies. This principle has again been illustrated using tea. Comparative Example 1 describes the analysis by high performance liquid chromatography (HPLC) of extracts of two different types of tea. The resulting chromatograms are shown in accompanying Figures 2 and 3. One experiment used untreated samples (chromatogram A in each of Figures 2 and 3) and the other used a treated sample (chromatogram B in each Figure).

The Figures show first of all that HPLC does not clearly distinguish between treated and untreated samples of tea since chromatograms 2 A and 2B are virtually identical, as are chromatograms 3A and 3B. Second, the Figures show that HPLC does not have the power to discriminate between different types of tea since the chromatograms of Figure 2 are virtually identical to those of Figure 3. In both these respects HPLC contrasts with the NMR spectroscopy/pattern recognition technique used in the present invention as illustrated in Example 2 and the accompanying Figures 4A and 4B. Many finished commercial tobacco products contain blends of two or more different types of tobacco. For instance, as mentioned above, cigarettes may consist principally of flue-cured tobacco with a proportion of Turkish tobacco added. Similarly pipe tobacco is frequently a blend of flue-cured and air-cured tobacco leaf. As discussed above, the NMR spectroscopy and pattern recognition technique of the present invention can be used in the standardization and differentiation of blends of tobacco plant material. The technique may thus be particularly useful in distinguishing between batches of of a given tobacco blend, and between different blends of tobacco. The process of the invention thus allows tobacco blends to be differentiated and standardized. The principle of applying the invention to blends of tobacco plant material is illustrated in Example 5, which describes how samples of a Traditional Chinese Medicine remedy, which is a mixture of plant materials, may be differentiated in accordance with the invention.

In the process of the invention the tobacco plant material typically consists of, or is derived from, a whole tobacco plant, a part of a tobacco plant, a tobacco plant extract, a tobacco plant fraction or a finished commercial tobacco product such as cigarettes, cigars, pipe tobacco or snuff.

The principles of the invention will be further illustrated in the following Reference Examples and Example with reference to the accompanying Figures, in which: Figure 1 is a principal component analysis (PCA) score plot of factor 3 (y axis) against factor 2 (x axis) for six samples of Panax ginseng obtained from different suppliers as described in Reference Example 1 . The symbols used in the figure are as follows: * = supplier 1; • = supplier 2; + = supplier 3; V = supplier 4; □ = supplier 5; and ® = supplier 6.

Figure 2 shows two HPLC chromatograms for Kemmun tea, using a system optimised for the separation of catechins as described in Comparative Example 1, in which A represents an untreated tea sample and B represents a tea sample which has been previously treated by passage through a solid phase extraction column. Figure 3 shows two HPLC chromatograms for Lapsang Souchong tea, using a system optimised for the separation of catechins as described in Comparative Example 1, in which A represents an untreated tea sample and B represents a tea sample which has been previously treated by passage through a solid phase extraction column. Figures 4A and 4B are PCA score plots of factor 2 (y axis) against factor 1 (x axis) for six different types of tea, obtained in Example 2 which follows. The symbols used in the figure for each type of tea are as follows: □ = Oolong; + = Kemmun; V = Lapsang Souchong; • = Darjeeling; * = Gunpowder; ♦ = Assam.

Figure.5 is a PCA score plot of factor 3 (y axis) against factor 1 (x axis) for the five different commercial samples of Tanacetum parthenium (feverfew) capsules obtained in Example 3. The symbols used for each sample are as follows: ® = sample 1; * = sampled ; D = sample 3; • = sample 4; and V = sample 5.

Figure 6 is a PCA score plot of factor 2 (y axis) against factor 1 (x axis) for the seven samples of Tanacetum parthenium harvested at different intervals after planting out, as described in Example 4. The symbols used are as follows: • = sample TO; V = sample Tl ; ® = sample T2; * = sample T3; + = sample T4; □ = sample T5; and 4? = sample T6.

Figure 7 is a PCA score plot of factor 2 (y axis) against factor 1 (x axis) for the Traditional Chinese Medicine remedy analysed in Example 5. The symbols used for each sample are as follows: • = supplier 1; * = supplier 2; and + = supplier 3.

Reference Example 1: Use of NMR Spectroscopic and Multivariate Analysis to discriminate between sources of Panax ginseng

Preparation of extracts

Samples of white Ginseng were obtained from six different commercial suppliers. White Ginseng is derived from the root of Panax ginseng CA. Meyer. The root is put in boiling water briefly and then soaked in the sugar juice. It is subsequently exposed and dried in the sun. White Ginseng is also known as Sugar Ginseng.

The dried plant root material was ground to a fine powder using an "Illico" blender for five minutes. 1 g of the powder was mixed with 20mL of cold water and stirred continuously for 2 hours on a shaker at 150 rpm at room temperature (22 °C). The extract was filtered through Whatman No. 1 filter paper, the filtrate collected and freeze-dried overnight. lOmg of the sample was dissolved in 1 mL of deuterated water for NMR spectroscopic analysis. 800 μL were used for NMR spectroscopic analysis and samples were referenced internally to TSP at 0.00 ppm. Protocol for NMR spectroscopy

Hydrogen- 1 NMR spectra were recorded on a Bruker DRX 600 Spectrometer operating at 600.13 MHz for the proton frequency, fitted with a BEST flow probe.

Spectra were a result of 64 scans of 20 ppm sweep width and were collected into 49152 data points. Acquisition time per scan was 2.0 seconds. Prior to transformation, a line broadening of 0.3 Hz was applied and the spectra were Fourier transformed. Referencing was to TSP at 0.00 ppm. NMR spectra were analysed using AMIX software to reduce the spectra into "histograms" containing buckets of

0.4 ppm width.

Principal component analysis

The resulting file was opened in Excel where normalisation was performed using the sum of the entire spectrum. The data resulting from this was then subjected to multivariate analysis using the Pirouette software package. Unsupervised PCA was performed using mean centring (and autoscaling).

Results

The data were converted into points on a PCA score plot. This is Figure 1 attached. The figure clearly shows six distinct clusters of points attributable to each of the Panax ginseng samples, thereby illustrating that the technique provides a means of discrimination between ostensibly identical samples of a given plant material.

Comparative Example 1: HPLC analysis of extracts of tea

The teas used were Kemmun and Lapsang Souchong. Extracts of each for HPLC analysis were prepared as follows:

Untreated samples Dried commercial tea sample (50 g) was ground to a fine powder for approximately 1 minute using a Moulinex "Illico' blender, to produce a homogeneous sample. 5g of the powder was removed and placed in a 100 mL glass conical flask at room temperature (22°C). 50 mL of boiling distilled water was poured over the tea sample which was stirred with a plastic rod and left at room temperature for 30 minutes. The extract was filtered through a Whatman No. 1 filter paper, the filtrate collected and freeze dried overnight. 10 mg of the sample were dissolved in 1 mL of distilled water.

HPLC equipment: Hewlett Packard series II 1090 Liquid Chromatograph HPLC column: C18 RP Hypersil 5μ, 150 x 4.6 mm

HPLC mobile phase: Methanol: Water: Orthophosphoric acid 20:79.9:0.1 v/v, flow rate 0.9 ml/minute, absorption wavelength 21 Onm.

The resulting chromatograms are accompanying Figures 2A (Kemmun) and 3A (Lapsang Souchong).

Treated samples Dried sea sample (50g) was ground to a fine powder for approximately 1 minute using a Moulinex "Illico" blender, to produce a homogenous sample, 50 mL of boiling distilled water was poured over the tea sample which was stirred with a plastic rod and left at room temperature for 30 minutes. The extract was filtered through a Whatman No. 1 filter paper, the filtrate collected and freeze dried overnight. 10 mg of the sample was dissolved in 1 mL of distilled water which was run through a Reverse Phase (RP) Cl 8 Isolute^R column under vacuum. 20μL were used for the HPLC analysis, following the method described above for the untreated samples.

The resulting chromatograms are accompanying Figures 2B (Kemmun) and 3B (Lapsang Souchong). The Figures show that HPLC does not clearly distinguish between treated and untreated samples of tea since chromatograms 2 A and 2B are virtually identical, as are chromatograms 3 A and 3B. The Figures also show that HPLC does not have the power to discriminate between different types of tea since the chromatograms of Figure 2 are virtually identical to those of Figure 3. Reference Example 2: Use of NMR Spectroscopy and Multivariate analysis to discriminate between types of tea

The teas used were Gunpowder, Darjeeling, Kemmun, Assam, Lapsang Souchong and Oolong. Untreated and treated samples of these teas were prepared as described in Comparative Example 1.

Protocol for NMR spectroscopy and Principal Component Analysis

The 10 mg tea samples were dissolved in 1 mL of deuterated water (D₂O). 800 μL were used for NMR spectroscopic analysis and samples were referenced internally to TSP at O.OOppm. The procedure described in Reference Example 1 was carried out on each of the tea extracts.

Results The data were converted into points on two score plots. These are Figures 4 A and 4B attached. Figure 4B shows six distinct clusters of points attributable to each of the tea extracts following treatment, whilst Figure 4A shows that before treatment clustering was poor.

The discriminating power of the NMR spectroscopy /PCA technique is demonstrated, contrasting strongly with the inability of HPLC to distinguish tea types as shown in Comparative Example 1.

Reference Example 3 Use of NMR Spectroscopy and Multivariate Analysis to discriminate between commercial samples of a herbal remedy

Preparation of extracts

Commercially available capsules of feverfew (Tanacetum parthenium) were obtained from three different manufacturers. The products were ostensibly identical, being hard gelatin capsules containing a 'standardised' feverfew powder.

Five samples, each consisting of 10 capsules were taken. One sample was taken from manufacturer A. Two samples, each bearing the same batch number, were taken from manufacturer B. Two samples, each bearing a different batch number, were from manufacturer 3.

The capsule contents from each sample were placed in a Moulinex "Illico" blender and the material ground to a fine powder for two minutes to produce a homogeneous sample. Water extracts were prepared for NMR spectroscopic analysis by dissolving 500 mg of the powder in 50 mL cold water and stirring the extract for four hours on a shaker at 150 rpm at room temperature (22°C). The extract was filtered through Whatman No. 1 filter paper, the filtrate collected and freeze dried overnight.

Protocol for NMR spectroscopy and Principal Component Analysis

The procedure described in Reference Example 1 was carried out on each extract.

Results The data were converted into points on a score plot (mean-centred). This is

Figure 5 attached, in which each point corresponds to one capsule. The plot shows that samples 1, 4 and 5 gave good clustering that reflects consistency between individual capsules in those batches of products. Samples 2 and 3 gave poorer clustering, reflecting greater variability in those particular products from capsule to capsule. The plot also highlights distinct differences between different batches of one product from one manufacturer (samples 4 and 5) which are claimed to be identical.

Reference Example 4: Use of NMR spectroscopy and Principal

Component Analysis to discriminate between samples of Tanacetum parthenium of differing maturity

Cultivation and harvesting of plant samples

The medicinal plant species Tanacetum parthenium (Feverfew) was grown in the UK from warranted seed obtained from CN. Seeds, Ely, UK. Seeds were sown on 18^th April and raised under glass. Plantlets emerged on 25^th April and were planted out on 24^th May in plots laid out on a grid basis to provide a randomised sampling regime.

Five samples consisting of 4 plants each were harvested from the plots for analysis at different intervals after planting out. Two additional samples were taken at final harvesting, stored frozen and analysed at one and two month intervals post- harvest. The dates of harvesting of the samples were as follows:

Preparation of extracts

Dried plant material was collected and immediately placed in a freezer at -20°C. After freezing, the material was placed in a freeze drier for 12 hours after which samples were checked to ensure they were dry. The material was placed in a Moulinex "Illico" blender, and the material ground to a fine powder for two minutes to produce a homogeneous sample. Water extracts were prepared for NMR spectroscopic analysis by dissolving 500 mg of plant material in 50 mL cold water and stirring the extract for four hours on a shaker at 150 rpm at room temperature

(22°C). The extract was filtered through Whatman No. 1 filter paper, the filtrate collected and freeze dried overnight.

Protocol for NMR spectroscopy and Principal Component Analysis lOmg of the sample as described above was dissolved in 1 mL of deuterated water for NMR spectroscopic analysis. 800 μL were used for NMR spectroscopic analysis and samples were referenced internally to TSP at 0.00 ppm. The procedure described in Reference Example 1 was followed.

Results

The data were converted into points on a PCA score plot (mean-centred) . This is Figure 6 attached. There is a clear clustering of points relating to the different samples. The results demonstrate that metabolic changes in a plant, for instance as manifested int the maturing process, can be represented on a PCA map. The process of the invention thus provides a means of discriminating between samples of a given plant material having different physiological states, in this case of differing maturity levels.

Reference Example 5: Use of NMR Spectroscopy and Multivariate

Analysis to distinguish between samples of a mixture of plant materials

The Traditional Chinese Medicine remedy known as Liu Wei Huang Wan, which contains six plant ingredients, was obtained from three different suppliers (denoted 1, 2 and 3). Preparation of extracts

100 tablets from each supplier were separated into groups of 10 and weighed. They were then crushed with a pestle and mortar and Ig of the resulting powder was transferred to a glass flask. Cold water (1 OOmL) was then poured over the powder and the mixture was refluxed for 30 minutes. The solution was cooled, filtered and freeze dried.

NMR spectroscopy and Principal Component Analysis lOmg of each sample prepared as described above was removed and dissolved in lmL of deuterated water. The procedure described in Reference Example 1 was carried out.

Results

The data were converted into points on a PCA score plot (mean-centred). This is Figure 7 attached. There is a clear clustering of points corresponding to samples from the three suppliers. These results demonstrate that the process of the invention can successfully discriminate between samples of multi-component plant materials which are ostensibly identical.

Example 1: Use of NMR Spectroscopic and Multivariate Analysis to discriminate between sources and batches of tobacco

Preparation of Extracts

The following eight samples of commercially available branded cigarettes were obtained:

10 cigarettes were removed from the pack of each of samples 1 to 8 and the tobacco material was removed from the paper. The tobacco material was then extracted by refluxing in 70% alcohol for 30 minutes. The resulting solution was filtered and the solvent removed by rotary evaporation and freeze drying.

Protocol for NMR spectroscopy and Principal Component Analysis lOmg of the residue of each sample as described above were taken and dissolved in 1 ml of perdeuterated methanol for NMR spectroscopic analysis. The procedure described in Reference Example 1 was then carried out on each of the eight samples.

Results The data were converted into points on a PCA score plot. There was clear clustering of points corresponding to each of the eight different samples. These results demonstrate that the process of the invention can successfully discriminate between tobacco from cigarettes sold under different brand names. The process can also distinguish different batches of a given commercial brand of cigarette (see samples 3 and 4, which are both "PrinslP" brand but have different batch numbers).

Application of the technique

A standard specification is established for each of the branded cigarettes on the basis of the NMR/PCA results obtained as described above. Subsequent samples of cigarettes sold under any of these brand names would then be tested for authenticity by submitting them to the protocol described above and comparing the results obtained with the relevant standard specification . By this means it is possible, for instance, to identify counterfeit cigarettes or branded cigarettes which have been adulterated with non-authentic tobacco material.

Claims

1. A process for establishing a standard specification for a tobacco plant material, the process comprising: (i) preparing a test solution or test extract of a sample of the tobacco plant material which is known to possess the or each property required for the standard: (ii) submitting the said solution or extract to analysis by a combination of NMR spectroscopy and a computer-based pattern recognition technique;

(iii) obtaining results from the analysis of step (ii); and (iv) defining a standard specification for the said tobacco plant material on the basis of the results obtained in step (iii).

2. A process for providing a sample of a tobacco plant material, which sample complies with a previously defined standard specification for that material, the process comprising:

(i') preparing a test solution or test extract of a candidate sample of the tobacco plant material; (ii') submitting the said solution or extract to analysis by a combination of

NMR spectroscopy and a computer-based pattern recognition technique; (iii') obtaining results from the analysis of step (ii'); and (iv') selecting the candidate sample if the results in step (iii') comply with the standard specification for the said tobacco material established in step (iv) of the process defined in claim 1.

3. A process for identifying, and rejecting, a sample of a tobacco material which fails to comply with a previously defined standard specification for that material, the process comprising:

NMR spectroscopy and a computer-based pattern recognition technique; (iii') obtaining results from the analysis of step (ii'); and

(iv') rejecting the candidate sample as sub-standard if the results obtained in step (iii') do not comply with the standard specification for the said material established in step (iv) of the process defined in claim 1.

4. A process according to claim 3 wherein the said candidate sample is a sample of a finished tobacco product which is suspected of being a counterfeit or of being adulterated with non-authentic tobacco material.

5. A process according to claim 4 wherein the tobacco product is a branded cigarette, cigar or pipe tobacco.

6. A process according to any one of the preceding claims wherein the said combination of NMR spectroscopy and a computer-based pattern recognition technique comprises: (a) submitting the test solution or test extract to NMR spectroscopy and recording one or more NMR spectra; and (b) submitting the data obtained from the or each NMR spectrum to a multivariate analysis to generate one or more points on a score plot.

7. A process according to any one of the preceding claims wherein the computer-based pattern recognition technique comprises principal component analysis (PCA).

8. A process according to any one of the preceding claims which further comprises purifying the test solution or test extract of the candidate sample prior to carrying out the combination of NMR spectroscopy and computer-based pattern recognition.

9. A process according to any one of the preceding claims wherein the tobacco plant material consists of, or is derived from, a whole tobacco plant, a part of a tobacco plant, a tobacco plant extract or a tobacco plant fraction.

10. A process for providing a standard specification for a tobacco plant material, the process comprising:

(i" ') preparing a test solution or test extract of a sample of the said tobacco plant material which is known to possess the or each property desired for the standard; (ii" ') submitting the test solution or test extract to NMR spectroscopy and recording one or more spectra; (iii" ') submitting the data obtained from the or each said NMR spectrum to a multivariate analysis to generate one or more points on a score plot; and (iv" ') defining a sphere of acceptability around the points generated in step (iii'") as the, or as part of the, standard specification for the said plant material.

11. A process according to claim 10 wherein the multivariate analysis of step (iii'") is performed using an unsupervised methodology.

12. A process for providing a sample of a tobacco plant material, which sample complies with a previously established standard specification for that material, the process comprising:

(i") preparing a test solution or test extract of a candidate sample of the said tobacco plant material; (ii") submitting the test solution or test extract to NMR spectroscopy and recording one or more NMR spectra;

(iii") submitting the data obtained from the or each said NMR spectrum to a multivariate analysis to generate one or more points on a score plot; and (iv") selecting the candidate sample as a sample which complies with the said standard specification only if the points generated on the score plot in step (iii") fall within a sphere of acceptability as defined in the standard specification established in step (iv'") of the process defined in claim 10.

13. A process according to claim 1 or 10 wherein the sample of the tobacco plant material which possesses said the or each property desired for the standard is a sample of authenticated or audited tobacco plant material of which the provenance is known.

14. A process according to claim 1 or 10 wherein the said sample of the tobacco material which possess the or each property desired for the standard is a sample of a specified tobacco cultivar, a sample of tobacco grown in a particular geographic location, a sample of tobacco of a specified leaf grade or a sample of tobacco which has been cured by a specified technique; or is a sample of a branded finished tobacco product.

15. A process according to any of the preceding claims wherein the tobacco plant material is derived from, or consists of, a blend of two or more different types of tobacco.