CORESTA joint experiment work in 2006 had compared data on a wide range of smoke constituents obtained from Kentucky reference cigarettes (1R5F and 2R4F), according to the existing methods used by participants. This work had identified that the methods used to determine aromatic amine yields in mainstream smoke would particularly benefit from further study to investigate the main weaknesses and influencing factors in their yield variability before progressing to full method standardisation. This report describes the output from a 2007 joint experiment to address these issues. Participating laboratories carried out experiments to investigate several factors that had been identified in the methodology as potential sources of variability. These were the amine derivative type, the derivatisation time and the point at which the addition of the internal standard for calibration occurred. A statistical assessment was made of their possible influence on aromatic amine smoke yields and yield reproducibility across different laboratories. Results showed that aromatic amines again had poor between-laboratory yield reproducibility. The stage at which the internal standard was added to the smoke sample had the most significant effect on yields. The least variable data were obtained when it was added directly after extraction from the filter pad rather than later in the process. It also appeared beneficial to use at least two calibration standards (i.e., an aminonaphthalene and an aminobiphenyl) to minimise yield differences although this recommendation was not supported by statistically significant data. Large differences in yields were not found when comparing the two studied derivatising agents especially when compared against the greater overall between-laboratory variability. Any differences between laboratories in total particulate matter and puff count at the smoke collection stage did not appear to significantly contribute to betweenlaboratory differences in yields. It appeared that some laboratories had significantly improved their methodology since the last study although high values for the between-laboratory reproducibility in this study were still found. It may be that significant improvements in reproducibility may not be forthcoming for compounds such as the aromatic amines measured at low nanogram smoke yields. Some important features that need to be controlled to minimise variability were identified in this study and will be incorporated within a collaborative study leading to a recommended method. Also, a wider range of product styles will need to be investigated, to determine the effects of differences in tobacco blends and product styles and the potential of greater product variability of commercial products. This should provide more robust estimates of within-laboratory repeatability and between-laboratory reproducibility.
Joint experimental work carried out in 2006 by the CORESTA Special Analytes Task Force compared yield data on a wide range of smoke constituents obtained from reference cigarettes according to the existing methods used by participants. This work identified that the methodologies that were used to determine yields of selected volatiles in mainstream smoke under the ISO smoking regime would benefit from further joint experiments. This report describes the output from the 2008 Joint Experiment on selected volatiles in smoke (1,3-butadiene, benzene, toluene, acrylonitrile, and isoprene). Its objectives were to investigate the main weaknesses and influencing factors in methodologies used by the participating laboratories and their effects on yield variability before deciding on one to take forward to a CORESTA recommended method. The Task Force considered this step was necessary before progressing to a full collaborative study using a recommended method. An experimental protocol was devised to investigate several factors such as the use of different calibration standards and the efficiencies of different trapping systems. The effects of other general factors identified from supplied methodology information as differing across laboratories were also analysed. A statistical assessment was made of their possible influence on smoke yields and yield reproducibility across different laboratories and is discussed in this report. Between-laboratory variability has been reduced since the last study indicating that some laboratories have improved their methodology although extremely high values for the among-laboratory variability were still found for acrylonitrile (> 100%) and 1,3-butadiene (~ 80%) when related to the mean yields. The means to reduce the variability in acrylonitrile and 1,3-butadiene yields are not apparent from the data and interpretations made in this study. However, when the different laboratories use the same methodology during the development of a recommended method at the next development stage then it is hoped that this high level of variability for acrylonitrile and 1,3-butadiene will be reduced to similar levels to those found for benzene, toluene, and isoprene. As in previous work, it was recognised that although a more intense smoking regime may be introduced into the regulatory arena in the future, it was decided that the current ISO smoking regime should be used for this joint experiment. A wider range of product styles will be investigated when the Task Force works towards a recommended method to take account of differing blends and designs and the potentially greater product variability of commercial products. This will provide robust estimates of within-laboratory repeatability and among-laboratory reproducibility and is intended to be reported in a later paper.
Three tobacco industry based laboratories determined selected mainstream components using their established in-house methods. Machine smoking was done according to the ISO smoking regime. The Test cigarettes smoked for this investigation were manufactured with different amounts of added glycerol, cocoa powder and sucrose. Variability between the three laboratories differed clearly for the analyzed smoke components. No overall effects due to the added ingredients on smoke components could be found. The high ‘tar’ products with the highest lodading of sucrose showed a slight increase in formaldehyde emissions among all three laboratories.
A recommended method has been developed and published by CORESTA, applicable to the quantification of selected carbonyl compounds (acetaldehyde, formaldehyde, acetone, acrolein, methyl ethyl ketone, crotonaldehyde, propionaldehyde and butyraldehyde) in cigarette mainstream smoke. The method involved smoke collection in impinger traps, derivatisation of carbonyls with 2,4-dinitrophenylhydrazine (DNPH), separation of carbonyl hydrazones by reversed phase high performance liquid chromatography and detection by ultra violet or diode array.
At the start of the process it was determined that most laboratories participating in the CORESTA Special Analytes Sub-Group (SASG) used a similar method involving such derivatisation and so this was chosen as the basis of the recommended method. Initial joint experiments, specific experiments by single laboratories and ongoing discussions addressed some methodological aspects that needed to be considered before moving to a recommended method.
As a first step, a joint experiment by 17 laboratories was carried out in 2009-2010 that investigated three features of the methodology on two reference cigarettes (3R4F and CM6) considered most important by SASG members. These were the volume of the impinger solution (25 or 35 mL); the type of mineral acid (perchloric or phosphoric) used to initiate the derivatisation and the time of derivatisation (5 or 30 min) before terminating the reaction with TrizmaTM base. Overall, it was concluded that these studied parameters in the methodology seemed to have little effect on the overall yield data, compared to the underlying variability among laboratories. The 25 mL impinger solutions appeared to give somewhat higher yields, although not with statistically significant differences, than those obtained when using 35 mL solutions.
Some laboratories volunteered to carry out other investigations, for example, to confirm the identity of both the Eand Z-isomeric acetaldehyde hydrazone peaks within the chromatogram of smoke carbonyls and to investigate methodology factors influencing the hydrazoneisomerisation.
The CORESTA recommended method (CRM) was produced through a final collaborative experiment involving 15 laboratories from 11 countries using 7 linear and 8 rotary smoking machines. Some notes are included in the CRM to inform other laboratories that might wish to adopt the method, concerning the main features that need to be well controlled to provide data as robust as possible and to provide similar repeatability and reproducibility data.
Statistical evaluations were made according to ISO 5725 recommendations and are included. As expected from previous work on other smoke components, the levels of reproducibility of carbonyl yields among laboratories are much greater than the levels found for “tar”, nicotine and carbon monoxide and given in the equivalent ISO standards. When expressing the reproducibility (R) value as a percentage of the mean yield among-laboratories and across all of the studied products, values ranged from 67-125% for formaldehyde; from 24-55% for acetaldehyde; from 41-108% for acetone; from 45-73% for acrolein; 31-75% for propionaldehyde; from 63-140% for crotonaldehyde; from 62-90% for 2-butanone and from 42-58% for butyraldehyde. The lowest “tar” yielding product gave the most variable data. These levels are generally in line with those determined for selected volatiles.
A recommended method has been developed and published by CORESTA, applicable to the quantification of selected volatiles (1,3-butadiene, isoprene, acrylonitrile, benzene, and toluene) in the gas phase of cigarette mainstream smoke. The method involved smoke collection in impinger traps and detection and measurement using gas chromatography/mass spectrometry techniques.
This report describes the final collaborative study applying the recommended method. It provides additional notes to inform other laboratories that might wish to adopt it, about some of the main features that need to be well controlled to provide data as robust and consistent as the data presented herein.
Data was provided by 15 industry-related and 4 independent laboratories and one governmental laboratory. Overall, 6 linear and 14 rotary smoking machines were used.
The joint experiments and collaborative work between the large number of participating laboratories has provided solutions to several methodological problems and reduced the high data variability that had initially been found particularly for 1,3-butadiene and acrylonitrile smoke yields.
Even so, the levels of reproducibility among laboratories are much greater than the levels found for ‘tar’, nicotine and carbon monoxide and given in the equivalent ISO standards. When expressing the reproducibility (R) value as a percentage of the mean yield among-laboratories and across all of the studied products, values ranged from 63-93% for 1,3-butadiene; from 36-62% for isoprene; from 41-110% for acrylonitrile; from 35-70% for benzene, and from 27-116% for toluene. For the higher ‘tar’ yielding products, the lower levels of variability were in line with those previously evaluated during Task Force work on standard methods for benzo[a]pyrene and tobacco specific nitrosamines. As expected, the lowest ‘tar’ yielding product gave the most variable data.
During 2012, three CORESTA Recommended Methods (CRMs) (1-3) were updated to include smoke yield and variability data under both ISO (4) and the Canadian Intense (CI) (5) smoking regimes. At that time, repeatability and reproducibility data under the CI regime on smoke analytes other than “tar”, nicotine and carbon monoxide (6) and tobacco-specific nitrosamines (TSNAs) (7) were not available in the public literature. The subsequent work involved the determination of the mainstream smoke yields of benzo[a]-pyrene, selected volatiles (benzene, toluene, 1,3-butadiene, isoprene, acrylonitrile), and selected carbonyls (acetaldehyde, formaldehyde, propionaldehyde, butyraldehyde, crotonaldehyde, acrolein, acetone and 2-butanone) in ten cigarette products followed by statistical analyses according to the ISO protocol (8). This paper provides some additional perspective on the data variability under the ISO and CI smoking regimes not given in the CRMs.
A CORESTA Recommended Method (CRM 75) has been developed and published, applicable to the quantification of tobacco-specific nitrosamines (TSNAs), namely, Nnitrosonornicotine (NNN), N-nitrosoanabasine (NAB), Nnitrosoanatabine (NAT) and 4-(N-nitrosomethylamino)-1-(3-pyridyl)-1-butanone (NNK) in cigarette mainstream smoke. The method involves smoke collection on a Cambridge filter pad under both ISO 3308 and the intense conditions adopted by Health Canada. An internal standard solution is added to the smoke collected on the pad and, after extraction, an aliquot is separated and quantitatively analysed by liquid chromatographytandem mass spectrometry (LC-MS/MS).
CRM 63 involving gas chromatography coupled with a thermal energy analyser (GC-TEA) was previously developed by the CORESTA Special Analytes Group that had been set up to develop recommended methods on smoke components. However, by 2009 most laboratories had moved to similar LC-MS/MS methods for TSNA analysis and so this technique was chosen as the basis of a new CRM and to complement CRM 63. Initial joint experiments, specific experiments by single laboratories and ongoing discussions identified methodological aspects that needed to be ‘standardised’ before moving to a CRM.
A joint experiment by 15 laboratories was carried out in 2010-2011 that investigated and identified important methodological features that needed to be controlled or clarified. CRM 75 was produced through a final collaborative experiment involving 20 laboratories from 12 countries using both linear and rotary smoking machines. Some notes are included in the CRM to inform other laboratories that might wish to adopt the method, concerning aspects that need to be well controlled to provide data as robust as possible and to provide similar repeatability and reproducibility data.
Statistical evaluations were made according to ISO 5725 guidelines and are included. Under ISO smoking, the levels of reproducibility (R) expressed as a percentage of the mean of TSNA yields across laboratories are much greater than the levels found for “tar”, nicotine and carbon monoxide and given in the relevant ISO standards. The R value was expressed as a percentage of the mean yield amonglaboratories and across all of the studied products. Under
ISO smoking R% values ranged from 25-60% for NNN; from 31-85% for NNK; from 47-58% for NAT and 40-99% for NAB. These levels are generally in line with those determined previously for TSNAs in CRM 63 and for other smoke analytes studied by the Special Analytes Group.
Under ‘intense’ smoking, R% values ranged from 30-88% for NNN; from 37-79% for NNK; from 47-83% for NAT and 42-111% for NAB. A plot of R against mean yields suggests that the ‘intense’ regime gives similar or slightly worse reproducibility than the ISO regime in spite of the higher yields generated.
Cigarettes with similar design features but with either cellulose acetate or dual carbon filters were made at 1-mg and 13-mg “tar” levels, as determined under the ISO smoking procedure. Products were smoked under the ISO, Massachusetts and Canadian smoking regimes to provide per-cigarette and per-puff yields of twelve vapour phase (VP) smoke components. The yields generated at the lit end of the cigarette and the significant yield reductions caused by filter ventilation, selective (carbon) adsorption, tobacco rod ventilation and diffusion were estimated in a modelling approach. For a “1-mg tar” carbon-filtered product it was estimated that the VP generated at the lit end was reduced by 99.4% to a machine yield of 17 µg/cig under ISO smoking conditions. Under the Canadian regime with 100% vent blocking, the estimated total VP was lowered 20% by tobacco rod effects and 15% by carbon filter adsorption giving a machine yield of 3487 µg/cig. The carbon filter adsorbed less efficiently partly due to the artificially high smoke temperatures through the filter that would probably not be tolerated by human smokers. Under the Massachusetts regime with 50% vent blocking, conditions better associated with human smoking, the total VP was lowered 51% by filter ventilation, 22% by tobacco rod effects and 17% by carbon filter adsorption giving a machine yield of 659 µg/cig. Ventilation is used to achieve “tar”/nicotine/carbon monoxide yield ceilings at 10/1/10 mg based on the current ISO smoking method. If future regulations were to mandate further reductions in VP then this will only be selectively achieved by increasing filter or tobacco rod ventilation/porosity or by using selective adsorption. It is inevitable that manufacturers will need to add further ventilation into their product to comply with such regulations and this should be reflected in any smoking regime. Furthermore, regimes with 100% vent blocking, that do not produce data reflecting the significant reductions in VP yields, provided to the smoker by ventilation, are misleading and their results will not correlate with relevant biomarker data. When proposing a different smoking regime, it is essential to understand the generation and transfer of smoke within cigarettes and factors involved in the subsequent data interpretation as described in this work. For regulatory evaluation purposes, cigarette characterisation using a regime that removes ventilation, one of the main design tools, is more misleading than the current ISO regime or one with partial vent blocking.
Regulatory authorities are currently discussing the measurement of and imposition of ceilings on certain smoke analytes, the so called ‘Hoffmann analytes’. However, as a pre-requisite, the measurement methods and the tolerances around the measurements first need to be established.
In 1999, the Cooperation Centre for Scientific Research Relative to Tobacco (CORESTA) set up a Task Force ‘Special Analytes’ to deal with analytical methodology for measuring ‘Hoffmann analytes’ under International Standard (ISO) smoking and to work towards the standardisation of methods. This paper describes the output and conclusions from a 2005-2006 joint experiment made within the Task Force representing laboratories currently able to analyse these compounds. Data were obtained on most ‘Hoffmann analytes’ from reference cigarettes (2R4F and 1R5F), collecting data according to the existing methods used by the nineteen participating laboratories, in order to describe the within and among laboratory variability and to see which methods could most benefit from more rigorous standardisation work.
In some cases, the applied statistical analysis found that methods could not well differentiate the 1R5F and 2R4F cigarettes of differing ‘tar’ yield. This was explained, in part, by the broad range of methods used by the participating laboratories but also indicated that there were significant inadequacies in the choice of some methods or weaknesses in their application.
Results indicate that ‘Hoffmann analyte’ data are generally more variable both within and among laboratories than nicotine free dry particulate matter (NFDPM); nicotine and carbon monoxide due to their lower smoke yields. Accordingly, tolerances around methods adopted for regulatory purposes will need to be proportionately higher.
Methods for benzo[a]pyrene (B[a]P) and tobacco-specific nitrosamines (TSNAs), already taken to CORESTA recommended methods or ISO standardised methods through the efforts of this Task Force, give some of the most reproducible results, showing the value of this process. However, these data strongly suggest that even these analytes have much higher among-laboratory variability than for NFDPM, nicotine and CO and, based on the only two available one point in time studies, may need tolerances in the range of 35-45% for B[a]P and 26-55% for TSNAs, if they are to be measured for regulatory purposes.
The collected data is useful to participating laboratories for internal method validation and laboratory accreditation, and data comparisons with others allow laboratories to identify strengths and weaknesses in their current methods.
However, much work still needs to be carried out to take most of the methods towards standardisation. Although some fundamental differences or areas of concern around the methodology are discussed herein, they are not comprehensive and there may be others that need to be addressed before methods can be considered ready to take to a Recommended Method and/or to an ISO Standard. These methodological issues are being addressed in further CORESTA work within this Task Force. Smoke analytes with the highest variability found in this study and those analytes that are currently of highest regulatory interest are being prioritised and after further joint experiments, the results are intended to be published.