To ensure quality, consistency and supply security of its portfolio over time, a company may need to make changes that affect all or much of its portfolio of products. Among other requirements, prior to introducing a changed tobacco product into interstate commerce, FDA requires reporting of Harmful and Potentially Harmful Constituents (HPHC). In our proposed approach, instead of testing each product individually, we propose conducting a designed experiment of a subset of products that encompasses the major design characteristics of the manufacturer’s portfolio and use statistical modeling to determine the HPHC yield for the rest of the portfolio. Additionally, such a modeling approach could also potentially be used to generate supporting information for other premarket submissions such as Substantial Equivalence Reports.

To demonstrate feasibility, we used 30 representative products that cover the range of cigarette design and filler parameters of our entire portfolio. One set of 30 products was manufactured using current cigarette paper and another set using paper from an alternate supplier. The experiment was controlled to minimize product, manufacturing and analytical testing variations between the products with the two cigarette papers. Models were developed to correlate the HPHC yields of the changed product to yields of the control product. For model validation, 12 different products were randomly selected from the remaining products. The predicted yields from the model were compared with the measured yields. Model predictions were robust and differences between measured and predicted values were within the ISO repeatability limits, thereby demonstrating feasibility of our proposed approach.