Suresh Ralapati, batf/National Laboratory Center

Method Scope. Samples of smokeless tobacco are ground, made basic with sodium hydroxide and nicotine is partitioned into methyl t-butyl ether. Sample extracts are analyzed by gas chromatography to determine the nicotine concentration in the smokeless tobacco sample. As first drafted (101), the method also specifies that nicotine recovery studies with a “vegetable matrix” will be conducted with each set of samples.
Sample Preparation. The method recognizes that smokeless tobacco is produced in various physical forms. Plug tobacco, twist tobacco and loose tobaccos are ground to a 4 mm particle size. Dry snuff, wet snuff and wet snuff portion packs are analyzed without grinding, although the packaging material or “pouch” from the wet snuff portion pack is ground. For all tobacco types, a minimum sample size of 100 g is recommended to ensure a uniform and representative sample. Tobacco samples are analyzed “as is” without drying. After grinding, 1 g samples of tobacco are wetted with 2 N sodium hydroxide and nicotine is partitioned into 50 mL of methyl t-butyl ether while shaking the sample for two hours.
End Determination. Nicotine is determined by capillary gas chromatography via the split injection mode using an apolar capillary column (5% phenyl / 95% methyl silicone) and flame ionization detection. Quantitation is based on internal standard calibration employing quinoline as the internal standard. A typical separation is found in Figure 25.
Nicotine Reporting Requirements. Nicotine concentrations are reported as mg/g tobacco to an accuracy level of two decimal places, along with summary statistics from repeated measurements. In addition to the total nicotine content found in the tobacco, a quantity termed “free base nicotine” is estimated from a “tobacco pH” measurement and the Henderson-Hasselbach equation. Nicotine concentrations are not corrected for the moisture content of the tobacco, but tobacco moisture is determined separately and reported with the other values.


The Determination of Nicotine in Cigarettes and Smokeless Tobacco (State Regulation). In late 1996, legislation was passed by the Commonwealth of Massachusetts that mandated new testing and reporting regulations for cigarettes and smokeless tobacco products sold in Massachusetts (74). In accordance with that law, draft regulations were proposed and a public hearing was announced for early 1997 (107 - 110). Nicotine testing and reporting regulations were revised and promulgated in final form in August 1997 (75,111). The first nicotine testing results for cigarettes and smokeless tobacco were reported to the Massachusetts Department of Health on December 15, 1997 as required by the new regulations. Similar regulations were adopted the following year in Texas. The first testing results reported to the Texas Department of Public Health were initially required on December 1, 1998 (76), however the reporting date was subsequently extended into the next year in view of uncertainties surrounding the ability to complete all necessary testing by the earlier deadline and other unresolved regulatory issues regarding cigarette additives and ingredients.

One of the stated purposes of the Massachusetts testing regulations is to provide “cigarette nicotine yield ratings”. Those ratings are described as “a composite of information intended to show the range of nicotine that each cigarette brand can be expected to deliver to the average consumer….” The testing rationale adopted by the regulations assumes the premise that the range of smoke nicotine that each cigarette brand can be expected to deliver to the average consumer can be based on “machine testing parameters that seek to reflect actual smoking behavior.” It is also assumed in the regulation that cigarette design features such as filter ventilation and cigarette tobacco nicotine content, as well as potential responses to cigarette design features such as “smoke pH” can be used to predict average consumer smoke nicotine intake.

To comply with state regulations, smoke nicotine, filter ventilation, tobacco nicotine and “smoke pH” are determined for all cigarette brand styles sold within the state. While four different cigarette tests are conducted, the nicotine yield rating is based on results from a single test rather than a composite of information: smoke nicotine results obtained with an exaggerated smoking regimen.

Smokeless tobacco manufacturers are required to test smokeless tobacco samples to determine tobacco pH, tobacco moisture content and the amount of nicotine in smokeless tobacco. In addition to this information, manufacturers must report the percentage of “unionized (free) nicotine,” the “total unionized (free) nicotine” in units of mg/g and a “smokeless tobacco nicotine delivery” rating based on the latter quantity.

As an era of increased tobacco product regulation begins, the choice of a draft method that is intended for analysis of a specific sample matrix (smokeless tobacco) as a state regulatory standard is scientifically troublesome. The endnotes section of the draft CDC method for smokeless tobacco states, “The comments and notes listed below can be described as Good Laboratory Practice guidelines; they are described in detail in this protocol to ensure minimal interlaboratory variability in the determination of nicotine, total moisture, and pH in smokeless tobacco [emphasis added].” Although the intended scope of application for draft methodology seems clear, state regulations exceed the intended method scope and require that cigarette cut filler tobacco shall also be analyzed with the smokeless tobacco methodology. As regulation of tobacco products increases, development of testing requirements based on sound scientific principles of method validation and interlaboratory testing will be a challenge, especially given the expedient alternative available to regulators of requiring published research methods without regard for any potential technical limitations.

Method Development and Validation Issues. As discussed by Borgerding et al. (112), the major U. S. cigarette manufacturers expressed numerous concerns to the Massachusetts Department of Public Health during the comment period in early 1997 as the testing regulations were being developed (113,114). Central to the comments were three points. First, the underlying premise that such things as an “average smoker” and an “average smoke yield” exist and could provide useful information to individual consumers about their nicotine intake or “average nicotine intake” was a flawed concept. Analytical smoking machines (i.e., the 20-port linear or rotary smoking machines currently used for standard methods such as the FTC, ISO, CORESTA, and TIOJ methods) cannot replicate actual human smoking behavior. These smoking machines are designed to take uniformly shaped puffs of a precisely set volume and duration when puffs are taken at a constant frequency. Human smoking behavior, in contrast, is variable. Puffing topography is not constant (a) puff to puff as the smoker smokes a cigarette, (b) from cigarette to cigarette for a particular smoker, or (c) from individual to individual (115). Second, without the benefit of scientifically accepted method development, method validation and interlaboratory collaborative evaluation studies, the “testing methods” prescribed by the regulations would not necessarily produce comparable data from one laboratory to the next. Illustrations of the problems that can be and that were encountered follow. Third, the testing guidelines provided by the regulations did not provide sufficient technical detail with which to conduct meaningful experiments. The Massachusetts Department of Public Health incorporated some revisions to the proposed regulations, but did not address the major concerns of the manufacturers. The cigarette manufacturers were thus obliged to follow the prescribed test methods in order to report the required information by the deadline set by the state.

Table 21 compares the method development and validation process followed for the federal and state nicotine testing regulations. While neither regulation has fulfilled all of the steps generally expected in such a process, it is clear that the state regulations have implemented the draft methodology for smokeless tobacco as though it is an official, standardized analysis methodology, without any substantial method validation.

Problems that Can Occur Using Non-Standardized / Non-Official Methods. Testing conducted in 1997 to comply with state regulations provides some insight into the accuracy and precision of the tobacco nicotine method when applied to cigarette cut filler tobacco. For example, Figure 26 depicts method quality control data and summarizes results for two different tobacco samples during a three-month period. During that time, more than 60 determinations each of a burley tobacco and a typical American blend cut filler were conducted. Mean nicotine concentrations of 25.89 ± 1.25 and 14.93 ± 0.73 mg/g were found for the burley tobacco and American blend cut filler tobacco samples, respectively, based on the quality control data. Thus, the coefficient of variation for repeated analysis of each tobacco type during the three-month period was less then 5 (i.e., 5% RSD).

While the method quality control data in Figure 26 provide an indication of the method precision that can be obtained by a single laboratory, data found in Figures 27 and 28 reveal information regarding interlaboratory accuracy and precision. As part of the1997 state compliance testing, one major tobacco manufacturer chose to have its cigarette brand styles tested by an independent contract laboratory. Although comparison of results from two laboratories is much more limited in scope than the typical eight laboratory comparison required when a research method is validated as an official standard method (102), data found in Figures 27 and 28 are the only example available to date in which more than one laboratory has used the Massachusetts tobacco nicotine method to analyze the same cigarette samples.

Figure 27 compares tobacco nicotine analysis results determined by the corporate laboratory with the results obtained by the contract laboratory. Tobacco nicotine concentrations determined by the corporate laboratory ranged from approximately 9 – 14 mg/cig, while results from the contract laboratory were approximately 8 – 13 mg/cig. Individual cigarette brand style results when compared from the two laboratories differ by as little as a few tenths of a milligram per cigarette, and as much as roughly two milligrams per cigarette. Regression analysis (Figure 28) suggests that results from the two laboratories are correlated, although there is both an offset (~ 1 mg/cig) and a bias (~ 19%) between the data sets. It is important to note that two separate laboratories following the same written protocol for identical samples obtained these results. Without a formal method validation process and an interlaboratory comparison study with a representative number of laboratories, it is not possible to assess which set of results (the corporate laboratory or the contract laboratory) are accurate. In fact, it is possible that neither set of results is correct.

The 1997 state compliance testing results also afford the opportunity to compare the results from two different tobacco nicotine methods when the methods are applied in the same laboratory. During 1997, analyses were conducted by the corporate laboratory using the Massachusetts tobacco nicotine method and a continuous-flow analysis method. The continuous-flow method is the standard method applied in the corporate laboratory. This method employs a colorimetric endpoint, rather than gas chromatography (34). As such, the method is not specific for the determination of nicotine; rather total alkaloids are measured and reported as nicotine.

Figure 29 compares the results from the two analytical methods for thirty-three cigarette brand styles. Tobacco nicotine concentrations determined by the Massachusetts tobacco nicotine method ranged from approximately 9 – 14 mg/cig, while results from the company standard method were approximately 9 – 15 mg/cig. Comparison of individual cigarette brand style results yields absolute differences between the two methods of roughly a few tenths of a milligram per cigarette to approximately 1.7 mg/cig. This range of difference is similar to that observed for the between-laboratory comparison with a single method. Regression analysis (Figure 30) suggests that results from the two methods are correlated, although there is both an offset (~ 1.6 mg/cig) and a bias (~ 17%) between the data sets. The continuous-flow method (standard method for company A), generally yields higher results consistent with the non-specific nature of the method, i.e. the nicotine results obtained represent a summary response of “all” tobacco alkaloids expressed as nicotine. Based on the type of data presented in Figures 29 and 30, the Massachusetts Department of Public Health has approved the continuous-flow method as an equivalent alternative for use in complying with nicotine testing and reporting requirements.

Should these be added to this section?

• 
  Include charts demonstrating that cigarette nicotine content not good predictor of smoke yield under standard or mass conditions
  
• 
  Describe advantages and disadvantages