Reader Comments

Post a new comment on this article

Comments from Ruth Dempsey, Chief Scientist Operations Philip Morris International, December 22, 2011, re: Wertz et al, 2011 “The Toxic Effects of Cigarette Additives. Philip Morris’ Project Mix Reconsidered: An Analysis of Documents Released through Litigation”, PLoS Medicine, Dec 2011, vol 8(12) (published online on December 21, 2011)

Posted by HeikeSchramke on 22 Dec 2011 at 14:39 GMT

Whilst I agree with Wertz et al that regulatory decisions on tobacco products must be based on sound science, I am of the opinion that their concerns on project MIX are unfounded. None of the suggestions raised by the PLoS article invalidate the conclusions of the project MIX studies. Nor do they undermine the weight of evidence constituted by decades of research on tobacco ingredients, which, together with epidemiological and observational data, has led many in the scientific and public health community to conclude that cigarettes with ingredients are not more toxic than cigarettes without ingredients.

We fully support a constructive scientific dialogue, hence would have appreciated if the authors would have contacted us to obtain the relevant data and actual study protocols instead of basing their incomplete analysis on documents they found on the Internet. As a matter of fact, we have been providing detailed toxicological data for several years to many regulators around the world and shared it with the interested scientific community.

In the absence of regulatory guidelines or internationally recognized methods for the toxicological assessment of tobacco ingredients, Philip Morris has developed its own internal practices based on well established toxicological principles that are similar to those used in other industries. The project MIX studies were performed according to such principles and guidelines, including OECD guidelines and INVITTOX protocols.

Project MIX was a large body of research performed in the late 1990’s to investigate the toxic effect of 333 ingredients used at the time in Philip Morris cigarettes. All the project MIX studies were reported, alongside the actual data, in four peer reviewed scientific publications in 2002 and their way of calculation was discussed in one of the papers (Rustemeier et al, 2002).

I would like to reassure Wertz et al, and other readers, on some points in particular.

Firstly, Wertz et al suggest that they find it difficult to understand where the list of 51 mainstream smoke constituents came from. These constituents, as stated in the second paper in the series (Rustemeier et al, 2002), corresponded to the standard list from the Consumer Product Safety Commission (CPSC, 1993), to which a number of the IARC group 2B, group 2A and group 1 carcinogens were added (IARC, 2004). The final list of 51 constituents was certainly state of the art and includes all the major classes of chemical constituents of cigarette smoke (PAHs, N-nitrosamines, phenols, aldehydes and metals as well). It is similar to other lists of tobacco constituents and includes, for example, most of the 18 of the WHO Tobreg constituents of concern and the constituents that were subsequently identified on the Health Canada list.

Secondly, Wertz et al further express concern on the way in which the data for the smoke constituents was presented. Representation of the results normalized to TPM was a conscious decision of the authors to best address the situation with the addition of the ingredients, and was openly discussed in the second paper (Rustemeier et al, 2002). Although the results could equally be analysed by normalization to nicotine or on a per cigarette basis, this would not have changed the overall findings, and would have been less representative because of the unequal contribution of the ingredients added at exaggerated concentrations to the test cigarettes. In addition all data for all 51 constituents on a per cigarette basis were published in Rustemeier et al (2002, Table 2). The data for nicotine for each group was also presented in this table which allows any interested scientist to calculate the data normalized to nicotine and to easily make alternative comparisons, as Wertz et al did.

The observed increase in TPM associated with the exaggerated loading of ingredients in these studies (as compared to actual use levels) is not automatically of toxicological importance. Ingredients are clearly different than tobacco, and many of them are volatile and transfer unchanged (see page 246 in Baker and Bishop, 2004). Thus the part of the TPM made of the unchanged ingredients is likely not to have many of the toxic components coming from the combustion processes of tobacco. This was also fully discussed in Rustemeier et al (2002).

Thirdly, Wertz et al are incorrect in suggesting that the size of the rat groups for the 90-day inhalation study was too small. As clearly described in the fourth paper in the series (Vanscheeuwijck et al, 2002), 268 rats in 6 groups were exposed to cigarette smoke from cigarettes with ingredients, 1 group without ingredients, and 1 group exposed to the smoke of a standard reference cigarette. Group size was up to 14 male and 14 female rats per group. This is entirely in accordance with the OECD guidelines for this assay (OECD, 1981). Furthermore the 90-day rat inhalation assay is a well established assay. Although lifetime carcinogenicity studies may be more appropriate in many circumstances, no lifetime inhalation study which models cigarette smoking induced lung cancers has yet been established. Indeed the changes observed in 90-day rat inhalation studies (or even during shorter periods of 28 days), are representative of longer studies, thus a 90-day inhalation study for cigarette smoke has been considered as standard practice by many scientists in the area (Dempsey et al., 2011).

Finally, Wertz et al take issue with the use of the Ames assay as a quantitative tool. This is also unfounded. Although the original assay as described by Maron and Ames (1983) was generally used for yes/no answers, it has since been developed and the use of the slope of the dose-response curves to make quantitative comparisons is well established (Edler, 1992; Roemer et al., 1998; Roemer et al., 2010). This was also referred to in the third paper in the series (Roemer et al, 2002). Importantly, the Ames assay is anyhow only one piece of the many findings provided by this series of paper. As is recognized practice in toxicological evaluation, researchers need to make a critical assessment of the overall evidence taking into account the whole range of experiments. In this series of studies, the overall weight of evidence suggests that ingredients have no impact on the inherent toxicity of tobacco smoke.

Other research that has been conducted over the last fifty years on the effects of tobacco ingredients on the chemical and biological properties of cigarette smoke also indicates that commonly used ingredients do not change the toxicity of smoke as measured in specified assays. The evidence provided by toxicological assays is further supported by observational data. This is an additional key piece of evidence in the toxicity assessment, corroborating the relevance of experimental findings to human smoking. This has led many in the scientific and public health communities to conclude that cigarettes with ingredients are not more harmful than cigarettes without ingredients. As an example, in 2011, a US Scientific Committee, TPSAC, having reviewed all the available evidence on menthol concluded that: “The evidence is insufficient to conclude that it is more likely than not that smokers of menthol cigarettes have increased risk for diseases caused by smoking compared with smokers of non-menthol cigarettes” (TPSAC, 2011).

Regulators and scientists who have relied upon the project MIX publications, (amongst many others reporting similar results), should be reassured that the findings that “these ingredients added to tobacco do not add significantly to the overall toxicity of cigarettes” are still valid.


Baker, R.R. and Bishop, L.J., 2004. The pyrolysis of tobacco ingredients. J. Anal. Appl. Pyrol. 71, 223-311 (p 246).
Dempsey, R., Coggins, C.R., and Roemer, E., 2011. Toxicological assessment of cigarette ingredients. Regul. Toxicol. Pharmacol. 61, 119-128.
Edler, L., 1992. Statistical methods for short-term tests in genetic toxicology: The first fifteen years. Mutation Research 277, 11-33.
IARC, 2004. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 83. Tobacco Smoke and Involuntary Smoking. International Agency for Research on Cancer, Lyon.
Maron, D.M. and Ames, B.N., 1983. Revised methods for the Salmonella mutagenicity test. Mutation Research 113, 173-215.
OECD, 1981. OECD guideline 413, subchronic inhalation toxicity: 90-day study. In: OECD Guidelines for testing of chemicals. Organisation for Economic Cooperation and Development, Paris.
Roemer, E., Meisgen, T.J., Tewes, F.J., and Solana, R.P., 1998. Discrimination of cigarette mainstream smoke condensates with the Salmonella reverse mutation assay. Toxicological Science 42, 295.
Roemer E., Tewes F.J., Meisgen T.J., Veltel D., and Carmines E.L., 2002. Evaluation of the potential effects of ingredients added to cigarettes. Part 3: In vitro genotoxicity and cytotoxicity. Food and Chemical Toxicology 40, 105–111.
Roemer, E., Wittke, S., Sticken, E.T., Piade, J.J., Bonk, T., and Schorp, M.K., 2010. The addition of cocoa, glycerol, and saccharose to the tobacco of cigarettes: Implications for smoke chemistry, in vitro cytotoxicity, mutagenicity and further endpoints. Beiträge zur Tabakforschung International 24, 117-138.
Rustemeier, K., Stabbert, R., Haussmann, H.J., Roemer, E., and Carmines E.L., 2002. Evaluation of the potential effects of ingredients added to cigarettes. Part 2: Chemical composition of mainstream smoke. Food and Chemical Toxicology 40, 93–104.
Tobacco Products Scientific Advisory Committee, 2011. Menthol Cigarettes and Public Health: Review of the Scientific Evidence and Recommendations.
U.S. Consumer Product Safety Commission in Consultation with the U.S. Department of Health and Human Services, 1993. Toxicity Testing Plan 5.
Vanscheeuwijck, P.M., Teredesai, A., Terpstra, P.M., Verbeeck, J., Kuhl, P., Gerstenberg, B., Gebel, S., and Carmines, E.L., 2002. Evaluation of the potential effects of ingredients added to cigarettes. Part 4: subchronic inhalation toxicity. Food and Chemical Toxicology 40, 113-131.
Wertz, M.S., Kyriss, T., Paranjape, S., and Glantz, S.A., 2011. The toxic effects of cigarette additives. Philip Morris' Project Mix reconsidered: An analysis of documents released through litigation. PLoS Med 8, e1001145.

Competing interests declared: Employee of Philip Morris International

Philip Morris' letter continues business as usual

glantz replied to HeikeSchramke on 04 Jan 2012 at 19:01 GMT

Philip Morris’ response sidesteps the two most important findings in our paper: (1) After it found that the levels of many toxins in the smoke from cigarettes with additives increased (and TPM, generally by even more), Philip Morris changed the Project MIX protocol to normalize by total particulate matter (TPM), thereby obscuring the absolute increases; and (2) The animal toxicology studies were seriously underpowered, making it unlikely that it would detect statistically significant changes in biological effects due to the additives.

Philip Morris defends the TPM normalization on the grounds that, “Although the results could equally be analysed by normalization to nicotine or on a per cigarette basis, this would not have changed the overall findings …” The data show otherwise. As reported in our paper, the per cigarette and per nicotine results show that 18 and 23, respectively, of the compounds in the smoke increased by 20% or more. In contrast, Philip Morris’ TPM normalization reduced these numbers to 6. Indeed, only by normalizing by TPM could Philip Morris conclude that, “a general reduction in most of the compounds can be observed (Rustemeier, et al., p.99, second paragraph).”

Philip Morris still does not explain why it selected the sample sizes it did for its animal toxicology studies. While it emphasized that there were “up to” 14 rats per group (most groups had 9 rats), it ignored the fact that the number of animals required for a study is determined by the desired power to detect an effect of specified magnitude. Philip Morris neither specifies the magnitude of the effects it was testing for nor the power of the studies it did to detect these specified effects. Indeed, our paper demonstrated that had the experiments in Project MIX been larger – even half the size of the experiments Philip Morris did to compare secondhand smoke to diesel exhaust -- it would likely have found many more statistically significant toxic effects of cigarette additives than it reported.

Philip Morris noted that it used standard well-accepted'experimental methods. We agree that the experiments seem well done. The problems with the conclusions that Philip Morris published are due to post-hoc normalizations and underpowered experiments that obscure the actual effects of the additives on smoke and biological toxicity.

The many other lesser details in Philip Morris’ letter are addressed below.

The material we used for our paper was hardly just based on some “documents [we] found on the internet.” As Philip Morris well knows, the documents that we used to prepare our paper were located in the UCSF Legacy Tobacco Documents Library (http://legacy.library.ucs...), a collection of over 70 million pages of previously secret tobacco industry documents that Philip Morris and the other major cigarette companies have been forced to make public through 2021 under a court order by the US Federal District Court in Washington, DC as part of the remedy phase of the US Department of Justice’s case, US vs Philip Morris USA, Inc., et al. under the Racketeer Influenced and Corrupt Organizations Act (RICO) (United States v. Philip Morris USA, Inc., 449 F. Supp. 2d 1, 941-44 (D.D.C. 2006), aff’d in part and vacated in part, 566 F.3d 1095 (D.C. Cir. 2009) (per curiam), cert. denied, 561 U.S. ___, 130 S. Ct. 3501 (2010), as amended by United States v. Philip Morris USA, Inc., No. 99-cv-2496 (D.D.C. Sept. 20, 2006), ECF No. 5765 (“Order #1021”).

The Court clearly stated its reasoning for requiring this continued disclosure:

"The Court finds that in order to prevent and restrain such RICO violations in the future, Defendants must create and maintain document depositories and websites
which provide the Government and the public with access to all industry documents disclosed in litigation from this date forward. Disclosing such information will allow the public to monitor what Defendants are doing internally and to assess the accuracy of future information it may make available about their activities and their products. Imposing such disclosure requirements will act as a powerful restraint on Defendants’ future fraudulent conduct (United States v. Philip Morris USA, Inc., 449 F. Supp.2d at 928-29)."

The understanding of Project MIX provided in our paper illustrates exactly the sort of public assessment of the accuracy of tobacco company statements about their activities and products that the Court anticipated in requiring continuing release of industry documents.

Philip Morris highlights the fact that the Project MIX results were published in a “peer reviewed” journal. While the journal Food and Chemical Toxicology, where Philip Morris published all four Project MIX papers does use peer review, it was hardly independent of Philip Morris. As reported in our paper, the then-editor of Food and Chemical Toxicology was a member of the US tobacco industry's Council for Tobacco Research Scientific Advisory Board and Philip Morris Scientific Advisory Board and had a long history of doing contract research and consulting for Philip Morris and 13 other editors and members of the editorial board had ties to the tobacco industry.

As noted in the paper, Edward Carmines, director of Project MIX, emailed coworkers that the review process was “an inside job. We went to a journal whose editor knew us.” (http://legacy.library.ucs...).

Philip Morris’ statement that “all the Project MIX studies were reported, alongside the actual data” is incorrect. As noted in our paper Project MIX tested at least one crucially important compound in cigarette smoke – ammonia – but did not report the results.

Our paper stated that we did not understand how Philip Morris selected the 51 compounds it analyzed. Philip Morris responded that these compounds “corresponded to” a standard list from the Consumer Product Safety Commission together with some International Agency for Research on Cancer carcinogens. As our paper noted, combining these lists yields 118 compounds. The unanswered question remains not the source lists Philip Morris used to identify candidate compounds, but rather why it excluded over half of the carcinogens and other toxins identified on its source lists.

Philip Morris’ letter asserts that the cigarettes it made for Project MIX used an “exaggerated loading of ingredients [additives] in these studies (as compared to actual use levels).” This statement directly contradicts what Philip Morris said in the Materials and Methods section (Rustemeier et al, p.94) of the published Project MIX paper: “Each ingredient was introduced at a low and a high level. The low levels approximate typical use levels considered to be reflective of those used in modern cigarettes … [emphasis added].”

Philip Morris continues to cite old papers as well as some of its own work (Roemer) to defend using the Ames test for dose-response. It does not address the problems of doing so discussed in references [138] to [141] of our paper.

Philip Morris’ letter attempts to support its assertion that additives “do not change the toxicity of smoke” by quoting the Food and Drug Administration Tobacco Products Scientific Advisory Committee’s statement that “the evidence is insufficient to conclude” that menthol increases disease risk. Saying that evidence available at a particular time is “insufficient to conclude” an effect is very different from concluding, as Philip Morris did in the Project MIX papers, that “these ingredients added to tobacco do not add significantly to the overall toxicity of cigarettes (Carmines, p. 77 and 90).”

Quite the contrary, as we concluded in our paper, the appropriate interpretation of the Project MIX results is that “many of the toxins in cigarette smoke increase substantially when additives are put in cigarettes, including the level of TPM, and that, assuming that the toxicological results from Project MIX represent unbiased estimates of the true biological effects, these data show many adverse biological consequences (and that the failure to reach statistical significance was the result of underpowered studies rather than lack of an effect).”

For the reasons discussed above and in our paper, we stand by our conclusion that, “In particular, regulatory authorities, including the FDA and similar agencies elsewhere who are implementing FCTC articles 9–11, could use the Project MIX data to eliminate the use of these 333 additives (including menthol, which is the major component of ingredient group 3) in cigarettes.”

Indeed, this exchange reinforces our recommendation that scientists and regulators cannot take at face value statements regarding the safety of its products offered by Philip Morris (or other tobacco companies).

Marcia Wertz, PhD RN
Thomas Kyriss, MD
Sumin Paranjape, PhD
Stanton A. Glantz, PhD

Competing interests declared: See statement in original paper.