Evaluating Analytical Methods for Characterization of Tobacco Products

The Family Smoking Prevention and Tobacco Control Act requires premarket approval of all new tobacco products commercially marketed after Feb. 15, 2007—unless they are shown to be “substantially equivalent” to an existing tobacco product. The act defines "substantially equivalent" as “having the same characteristics or having different characteristics but not raising different questions of public health.”

The high cost of conducting exposure studies for premarket approval gives tobacco companies a significant incentive to demonstrate that newly marketed products are substantially equivalent to products already on the market. Regulators also need to have data that allows them to determine whether or not a new product is likely to introduce new and unique human health risks compared to other products that have been previously approved.

Products that are shown to be the same in design, heating source, materials, ingredients, composition and other key features as previously approved products may easily meet this threshold. However, in many cases more sophisticated analytical and statistical techniques are needed to demonstrate the equivalency of a new tobacco product. Battelle conducted internal research to validate analytical methods to be used for characterization of tobacco products for purposes of demonstrating Substantial Equivalence.

Tobacco products contain a complex mix of natural components and additives, including a large number of volatile and semi-volatile organic compounds at trace-level concentrations. This complexity makes characterization of these products highly challenging.

Battelle researchers tested a combination of two-dimensional gas chromatography with time-of-flight mass spectrometric detection (GC×GC-TOFMS) as a chemical analysis technique and a Balanced Random Forest (BRF) method as a statistical pattern recognition technique. GCxGC-TOFMS analysis has been shown to provide selective and highly sensitive detection of many organic compounds found in tobacco products. Random Forest methods are powerful techniques that can be used for sample attribution—especially for data sets with limited replicates and many variables. Using these techniques allows researchers to develop a chemical “fingerprint” of each tobacco product that can be compared statistically to demonstrate equivalence.

Researchers used four known tobacco products to conduct two separate two-product comparisons in order to validate the method. A quantitative approach was used to assess whether the organic contents of the two products appear to be equivalent at a specified confidence level.