Like the NAS panel in Clearing the Smoke,  LSRO believes that the primary,  if not the only objective of testing ingredients added to cigarettes should be to assure,  to the extent practical,  that ingredients do not increase the premature mortality and morbidity known to be associated with cigarette smoking.   Prospective epidemiological studies established this association with cigarette smoking.   However,  epidemiological associations alone do not prove causation.   Understanding causation requires diverse kinds of data.   For LSRO,  the descriptive standard for adverse health effects of cigarettes remains the long-term observation of human smokers.   Thus,  the best comparison would be between smokers of cigarettes containing an added ingredient with smokers of nearly identical cigarettes lacking this ingredient.   However,  observational studies such as these present many challenges.   For the foreseeable future,  the optimal public health goal will remain prevention of smoking initiation and encouragement of smoking cessation.

Manufacturers make cigarettes from tobacco,  a natural product.   Burning a cigarette produces smoke.   Normally a combustion aerosol,  like cigarette smoke,  consists of both particles and gases and also contains a mixture of many thousands of chemical substances,  some short-lived and reactive.   Cigarette smoke is dynamic in nature,  changing rapidly with time,  airflow,  puff frequency,  and type of cigarette.   Aerosol particles exchange substances with the gas phase,  and their chemical composition(s) vary over time.   When humans inhale cigarette smoke,  their mouths and respiratory tract surfaces exposed to the smoke react with,  and absorb,  many of the chemical substances,  sometimes producing effects at the site of deposition.   Once deposited,  these chemicals and/or their reactive products may be transported to systemic tissues by circulation.   Biological effects depend in part on complex interactions between the substances and/or their components and the target tissues.

Epidemiological studies have shown that certain major causes of death,  cancer,  cardiovascular diseases,  and chronic obstructive pulmonary disease (COPD),  contribute more than half of the premature mortality associated with cigarette smoking.   All of these diseases increase in incidence as unexposed nonsmokers age,  as smokers age,  as smokers consume more cigarettes,  and as smokers consume cigarettes for longer periods of time.   Among smokers,  latency and reversibility vary by disease.   For example,  the induction of lung cancer has a long period of latency,  and lung cancer incidence eventually decreases to nearly background rate after cessation of smoking.   In contrast,  COPD does not revert to background rate after cessation.

The testing of the adverse effects of ingredients added to cigarettes should evaluate adverse effects associated with (A) inhalation of the ingredient or its pyrolysis products within the hot smoke matrix,  (B) inhalation of cigarette smoke constituents altered by the ingredient,  and/or (C) changes in smoking behavior,  so that smoke inhalation pathways change.   Tests to predict these effects must have both internal validity (reliability,  reproducibility,  data quality,  and positive signal-to-noise characteristics) and external validity.   An invalid test cannot eliminate a false hypothesis.   Useful tests should generate data that also have external validity consistent with observed human outcomes.

In addition to observational studies of human smokers,  scientists generate cigarette smoke artificially using machines.   Test data reveal many interesting characteristics of the cigarette smoke matrix and the chemical substances in it,  but the results of these tests do not yield satisfactory explanations of the diseases seen in human smokers.   In addition,  few test data have addressed the effects of ingredients added to cigarettes.   Thus,  testing these ingredients will reveal significant research needs.

Data about the pyrolytic fate of an ingredient under cigarette smoking conditions would be uniformly desirable for all ingredients,  whether the ingredient transfers to smoke or,  pyrolyzes in whole or in part.   Research strategies will be necessary to support the testing of potential ingredient pyrolysis in burning cigarettes.   In determining the feasibility of testing ingredients,  the key is preparation of comparative test cigarettes,  nearly identical to a commercial product but lacking or containing lesser,  defined amounts of an ingredient.   Within the precision of standard methods,  test data about the effects of combustion within the cigarette also will reveal whether the ingredient shifts the chemical composition of other substances in the smoke matrix.

In evaluating the feasibility of testing ingredients added to cigarettes,  LSRO has concluded that the application of every conceivable test is neither necessary nor desirable.   For example,  while research into methods of pyrolysis and transfer during human smoking proceeds,  and while a scientific consensus about new methods arises,  LSRO recommends that investigators continue to include reference cigarettes,  such as those using the recipes of University of Kentucky for comparison,  and to use the reference smoke generation conditions specified by the Federal Trade Commission or International Standards Organization,  unless a compelling scientific rationale dictates otherwise (Davies & Vaught,  1990;   Federal Trade Commission,  1967a;   International Organization for Standardization,  2000).   Extensive comparative data are available for these reference cigarettes,  providing internal standards.

Research needs exist on the pyrolysis and transfer of ingredients under common cigarette smoking conditions.   There is a need to know,  for example,  whether transfer in smoke is concentration dependent.   Consensus methods for data acquisition about pyrolysis and transfer do not exist.   Research needs also exist about observation studies of changes in relative mortality and morbidity,  particularly for studies to achieve pre-specified limits of statistical power.   Consensus methods for data acquisition are not available when conducting studies of comparative mortality,  although smoking is an everyday activity.   In addition,  LSRO has not yet specified tests that would constitute a balanced (or optimal) set.   More research about general testing approaches would aid this task.

Because manufacturers already use many ingredients,  setting priorities on the evaluations of these ingredients makes good sense.   The need to set priorities explains why LSRO regards an incremental approach as beneficial.   A checklist of specific tests to apply to all added ingredients is neither feasible,  nor desirable.   A program of testing ingredients added to cigarettes will reduce the possibility of an additive-related adverse human health effect,  beyond the premature mortality and morbidity already seen with cigarette smoking.   In this regard,  manufacturers should retest their proprietary mixtures of added ingredients,  in the proportions used in their brands of cigarettes,  to self-assure that their mixtures lack effects,  which were not seen with individual ingredients,  generated through some interaction between ingredients.

Scientists can evaluate data from balanced,  diverse tests of an ingredient and identify a maximum level of the ingredient per cigarette,  based on these data,  or specify what additional testing would be required to identify the level.  

LSRO approaches the testing by estimating relative risk,  defined as the risk of exposure to a cigarette with the ingredient,  relative to the risk of exposure to a cigarette without the ingredient.   An "unchanged relative risk" does not imply that the underlying activity,  cigarette smoking,  is safer or that addition of the ingredient has changed the risk of cigarette smoking.   LSRO is not attempting to test ingredients in an effort to achieve safer cigarettes.   Relative risk depends on testing paired cigarettes that differ only in the presence or absence of an ingredient or mixture of ingredients.   The endpoint here is no change in risk.

Although the addition of ingredients to tobacco is unlikely to change significantly the adverse health effects of cigarettes based on the magnitude of the health effects of cigarettes and the incremental mass of pyrolyzed materials contributed by the added ingredients.   Guidelines for cigarette labeling and specification of excursion limits for ingredients added to cigarettes developed by an independent,  outside group of scientists will insure an objective and uniform interpretation of data and further minimize the possibility of a contribution to adverse health effects.   Overall,  LSRO concludes that testing of ingredients added to cigarettes with this objective is both feasible and worthwhile.