Chemical analysis in the regulation of
tobacco products
Chris Wright PhD, Derek Mariner PhD,
Mark Williams MBA & Chris Proctor PhD
British American Tobacco Group Research & Development, Southampton UK
68th Tobacco Science Research Conference, Charlottesville VA, 28 Sept – 1 Oct 2014
1
Overview
1. Tobacco product regulation
2. Regional requirements and substances of interest
3. Tobacco product emissions
4. Regional trends for selected HPHCs in smoke
5. Evaluating method performance
6. Ruggedness and how to achieve it
7. Demonstrating continuing improvement
8. Opportunities for shared learning
2
Abbreviated list of HPHCs in
Cigarette Smoke
Acetaldehyde
Acrolein
Acrylonitrile
4-Aminobiphenyl
2-Aminonaphthalene
Ammonia
Benzene
Benzo[a]pyrene
1,3-Butadiene
Carbon monoxide
Crotonaldehyde
Formaldehyde
Isoprene
Nicotine
NNK
NNN
Toluene
Abbreviated list of HPHCs in
Smokeless Tobacco
Acetaldehyde
Arsenic
Benzo[a]pyrene
Cadmium
Crotonaldehyde
Formaldehyde
Nicotine
NNK
NNN
2. Classes of constituents – FDA TPSAC April 2012
Trace metals
Arsenic
Beryllium
Cadmium
Chromium
Cobalt
Lead
Mercury
Nickel
Polonium-210
Selenium
Uranium-235
Uranium-238
Polycyclic Aromatics
Benz[a]anthracene
Benz[j]aceanthrylene
Benzo[b]fluoranthene
Benzo[k]fluoranthene
Benzo[b]furan
Benzo[a]pyrene
Benzo[c]phenanthrene
Chrysene
Cyclopenta[c,d]pyrene
Dibenz[a,h]anthracene
Dibenzo[a,e]pyrene
Dibenzo[a,h]pyrene
Dibenzo[a,i]pyrene
Dibenzo[a,l]pyrene
Indeno[1,2,3-cd]pyrene
5-Methylchrysene
Naphthalene
Alkenes /
cyanoalkanes
1,3-Butadiene
Isoprene
Hydrogen cyanide
Nitromethane
Others
Aromatic amines
Aromatics
4-Aminobiphenyl
1-Aminonaphthalene
2-Aminonaphthalene
o-Anisidine
2,6-Dimethylaniline
o-Toluidine
Benzene
Ethylbenzene
Toluene
Caffeic acid
Catechol
o-, m-, p-cresol
Phenol
Aflatoxin B1
Coumarin
Nitrobenzene
2-Nitropropane
Carbon monoxide
Chlorinated dioxins /
furans
Acetaldehyde
Acetone
Acrolein
Crotonaldehyde
Formaldehyde
Methyl ethyl ketone
Propionaldehyde
Heterocycles
Quinoline
Styrene
Heterocyclic aromatic amines
Alkaloids
Anabasine
Nicotine
Nornicotine
Low MW
volatiles
Acetamide
Acrylamide
Acrylonitrile
Ammonia
Ethyl carbamate
(urethane)
Ethylene oxide
Furan
Hydrazine
Propylene oxide
Vinyl acetate
Vinyl chloride
Carbonyls
N-nitrosamines
4-(Methylnitrosamino)-1-(3-pyridyl)-1butanone
N-Nitrosodiethanolamine (NDELA)
N-Nitrosodiethylamine
N-Nitrosodimethylamine (NDMA)
N-Nitrosomethylethylamine
N-Nitrosomorpholine (NMOR)
N-Nitrosonornicotine (NNN)
N-Nitrosopiperidine (NPIP)
N-Nitrosopyrrolidine (NPYR)
N-Nitrososarcosine (NSAR)
A-α-C (2-Amino-9H-pyrido[2,3-b]indole)
Glu-P-1 (2-Amino-6-methyldipyrido[1,2a:3',2'-d]imidazole)
Glu-P-2 (2-Aminodipyrido[1,2-a:3',2'd]imidazole) CA
IQ (2-Amino-3-methylimidazo[4,5f]quinoline)
MeA-α-C (2-Amino-3-methyl)-9Hpyrido[2,3-b]indole)
PhIP (2-Amino-1-methyl-6phenylimidazo[4,5-b]pyridine)
Trp-P-1 (3-Amino-1,4-dimethyl-5Hpyrido[4,3-b]indole)
Trp-P-2 (1-Methyl-3-amino-5Hpyrido[4,3-b]indole )
3. Constituents of emissions from novel tobacco products
Substances associated with alternative nicotine / tobacco product emissions
3R4F MS smoke (whole)
Disposable E-cigarette aerosol
Single puff from each product
analysed using thermal
desorption, capillary gas
chromatography, TOF MS
Heated Tobacco aerosol
5
3. Emissions from Electronic Cigarettes
Electronic cigarette aerosols contain very low levels of most HPHCs
Substances other than HPHCs may be present in e-cigarette
aerosol, depending upon the e-liquid and device.
But are they present at levels significant to health?
1,3-Dioxan-5-ol
cis-Aconitic anhydride
Maleic anhydride
1-Cyanovinyl acetate
Citraconic anhydride
Malic acid
4-Methylbenzophenone
Menthane (p-menthane)
Citric acid
Abieta-8,11,13-trien-18 -oic acid Decanedioic acid, bis (2-ethylhexyl) ester
Menthol
(dehydroabietic acid)
Menthone
Ethanol
Acetaldehyde
Ethyl acetate
Methyl 2-benzoylbenzoate
Acetic acid
Methyl Acetate
Formaldehyde
Acetic anhydride
Glycerol
Methylglyoxal
Acrolein
Methyl Vinyl Ketone
Glycidol
β-nicotyrine
Glyoxal
Nicotine
Benzene
Hexadecanoic acid
Octadecanoic acid (stearic acid)
Benzo[a]pyrene
(palmitic acid)
Phenol
Bisphenol A
Isoprene
Propanoic acid
Butyraldehyde
Itaconic anhydride
Toluene
Red = HPHC reported in e-cigarette aerosol
6
3. 3R4F smoke vs E-cigarette aerosol
HCI, comparison of selected constituents, scaled by consumption (14 cigarettes vs 300 puffs)
5000
4500
4000
3500
3000
2500
2000
1500
3R4F
E-cig
1000
500
0
7
3. Emissions from tobacco heating products
Heated tobacco emissions are different to cigarette smoke
Tobacco heating product emissions contain lower levels of HPHCs than
cigarette smoke.
They also contain other substances (as does cigarette smoke).
1,3-Butadiene
1-aminonaphthalene
2-Aminonaphthalene
3-Aminobiphenyl
4-Aminobiphenyl
Acetaldehyde
Acetone
Acrolein
Acrylonitrile
Ammonia
Benzene
Benzo[a]pyrene
Butyraldehyde
Carbon monoxide
Catechol
Crotonaldehyde
Formaldehyde
Hydrogen Cyanide
Hydroquinone
Isoprene
Methyl ethyl ketone
NAB
NAT
Nicotine
Nitric Oxide
NNK
NNN
o-Cresol
m+p-Cresol
Phenol
Propionaldehyde
Pyridine
Quinoline
Resorcinol
Styrene
Toluene
Red = detected in heated tobacco aerosol
Grey = awaiting data
8
3. 3R4F smoke vs heated tobacco aerosol vs E-cigarette vapour
HCI comparison of selected constituents per cig / consumable / 10 puffs
400.0
350.0
300.0
250.0
200.0
3R4F
150.0
HnB
E-Cig
100.0
50.0
0.0
9
4. Regional trends in selected smoke constituents
How similar are distributions in different regions?
Product testing results can be presented and interpreted in
different ways. Data must be fit for intended use, and this is
influenced by analytical sensitivity, accuracy and precision,
especially if data are provided by multiple laboratories
The following slides present ‘snapshot data’ for BaP, NNN
and 1,3-Butadiene in mainstream smoke from commercial
products in different regional markets
Do different regions have similar ranges of concentration?
Would common ‘global’ methods be applicable?
10
4. N-nitrosonornicotine in mainstream smoke, HCI
600
Market A
500
500
400
400
NNN (ng/cig)
NNN (ng/cig)
600
300
200
100
0
300
200
100
0
Product Index
600
Market C
500
500
400
400
NNN (ng/cig)
NNN (ng/cig)
600
300
200
100
0
Market B
Product Index
Market D
300
200
100
Product Index
0
Product Index
Graphs show: Mean Min Max, of replicate measurements by product and median
Data provided by Alison Eldridge,
BAT Group Research & Development.
11
4. 1,3-Butadiene in mainstream smoke, HCI
160
160
Market A
120
100
80
60
40
120
100
80
60
40
20
20
Product Index
160
Product Index
160
Market C
1,3-Butadiene (µg/cig)
120
100
80
60
120
100
80
60
40
40
20
Market D
140
140
1,3-Butadiene (µg/cig)
Market B
140
1,3-Butadiene (µg/cig)
1,3-Butadiene (µg/cig)
140
Product Index
20
Product Index
Graphs show: Mean Min Max, of replicate measurements by product and median
Data provided by Alison Eldridge,
BAT Group Research & Development.
12
4. Benzo[a]pyrene in mainstream smoke, HCI
50
50
Market A
Market B
40
B[a]P (ng/cig)
B[a]P (ng/cig)
40
30
20
20
10
10
0
30
0
Product Index
50
50
Market C
40
B[a]P (ng/cig)
B[a]P (ng/cig)
40
30
20
10
0
Product Index
Market D
30
20
10
Product Index
0
Product Index
Graphs show: Mean Min Max, of replicate measurements by product and median
Data provided by Alison Eldridge,
BAT Group Research & Development.
13
4. Method stability 1,3-Butadiene
Laboratory Control Cigarette, 3R4F
160
1,3-Butadiene (µg/cig)
140
120
100
80
60
40
20
AUG10
SEP10
OCT10
NOV10
DEC10
JAN11
FEB11
MAR11
APR11
MAY11
Month
14
4. Method stability Benzo[a]pyrene
Laboratory Control Cigarette, 3R4F
50
B[a]P (ng/cig)
40
30
20
10
0
AUG10
SEP10
OCT10
NOV10
DEC10
JAN11
FEB11
MAR11
APR11
MAY11
Month
15
5. Evaluating method performance
Specifying method performance requirements
Market
Toxicant
Units
N
Mean
StdDev
Min
P25
Median
P75
Max
A
1,3-Butadiene
µg/cig
925
106.9
14.2
63.3
97.4
107.3
116.4
147.3
A
B[a]P
ng/cig
925
18.0
4.6
8.2
14.8
17.7
20.7
43.2
A
NNN
ng/cig
925
68.0
56.8
0.0
31.2
51.2
79.4
376.5
B
1,3-Butadiene
µg/cig
695
70.8
16.2
31.4
60.3
68.1
79.3
126.3
B
B[a]P
ng/cig
700
18.6
4.9
7.1
15.2
18.1
21.3
42.2
B
NNN
ng/cig
700
139.1
52.4
31.0
104.0
135.5
163.0
423.0
C
1,3-Butadiene
µg/cig
840
97.1
15.9
50.0
88.4
99.1
108.0
140.9
C
B[a]P
ng/cig
840
20.2
5.4
10.4
16.2
19.4
23.7
38.5
C
NNN
ng/cig
840
142.5
69.5
12.6
101.5
132.8
173.0
477.7
D
1,3-Butadiene
µg/cig
700
98.8
20.5
51.8
83.1
101.3
112.9
156.5
D
B[a]P
ng/cig
700
13.1
3.9
5.1
10.4
12.7
15.3
30.3
D
NNN
ng/cig
700
150.4
78.1
33.2
102.9
133.4
174.4
515.2
16
5. Method performance requirement – 1,3-butadiene
160
Market B
140
1,3-Butadiene (µg/cig)
120
100
80
75th percentile
Market B median
60
40
20
Product Index
To exclude with 95% confidence all products above the 75th percentile would
require a method reproducibility SD (sR) ≤ 3.99µg/cig
Method reproducibility limit = (79.3 – 68.1) µg/cig;
[1.96√2.sR] = 11.2µg/cig;
sR = 3.99µg/cig;
17
5. Performance of current methods
CORESTA Recommended Method No 70
Determination of selected volatile organic compounds in mainstream cigarette
smoke by GC-MS, July 2014
For a 10mg delivery product under HCI conditions, sR = 10.78µg/cig
Method reproducibility does not support the desired level of discrimination.
Mean = 73.77µg/cig; R = 30.27µg/cig; sR = 10.78µg/cig
5. Method performance requirement – NNN
600
Market D
500
NNN (ng/cig)
400
300
200
75th percentile
Market D median
100
0
Product Index
To exclude with 95% confidence all products above the 75th percentile would
require sR ≤ 14.60ng/cig.
R = 41ng/cig; [1.96√2sR] = 41ng/cig;
sR = 14.60ng/cig
19
5. Performance of current methods
CORESTA Recommended Method No 75
Determination of tobacco-specific nitrosamines in mainstream cigarette smoke by LC-MS/MS,
July 2014
sR = 12.7ng/cig : Reproducibility supports the desired level of discrimination.
If a decision is based upon a ratio (e.g. analyte/nicotine), the reproducibility of
measurement of nicotine must also be considered.
Sample 2 Mean = 87.5ng/cig;
R = 35.9ng/cig; sR = 12.7ng/cig
6. Rugged test methods
What makes a method able to tolerate minor changes without impacting performance?
Measurement by ratio – use a suitable
internal standard (IS)
If low signal/noise, increase the selectivity
of measurement
- improve chromatographic separation;
- use a more selective detection
technique (e.g. MS, MS/MS)
- remove matrix artefacts (cleanup).
For mass selective detection, use stable
isotope dilution. Consider the number of
unique ions required to assure adequate
confidence in chemical identity
Apply system suitability criteria for
selectivity and signal/noise
Consider technical guidance and data
acceptance criteria applied to similar
methods in other regulated industries
21
7. Demonstrating continuing improvement
Interlaboratory studies can meet different needs
Method performance studies, such as those conducted by CORESTA,
enumerate the precision of measurement. In the absence of Certified
Reference Materials, it is difficult to assess accuracy of measurement
Proficiency studies conducted in
accordance with ISO/IEC 17043 and ISO
13528 evaluate the performance of
participating laboratories against preestablished criteria by means of
interlaboratory comparison. These include
quantitative schemes for the
measurement of constituents in test items.
22
7. Proficiency studies
Can facilitate improved agreement of results
For example, by applying a target standard deviation based on fitness for purpose and
by collating information on methods to share best practice.
2007-2008 Annual Asia Collaborative Study on ISO Tar & Nicotine
23
7. Predicting achievable performance
The Horwitz function
Target Reproducibility [ RSDR ] can be calculated from the Horwitz function.
The Horwitz function is based upon observations (drawn from thousands of collaborative
studies) of the relationship between analyte concentration and reproducibility.
PRSDR (%) = 2 C-0.1505
HorRat = RSDR/PRSDR
Reproduced from AMC technical Brief 17,
Royal Society of Chemistry 2004.
24
8. Opportunities for shared learning
Best technical practice, internal and external quality assurance
Proficiency studies
-
represent full range of product variation in the population to be tested;
-
apply standards consistent with best practice (e.g. target SD);
-
address substances of technical priority (e.g. of regulatory interest);
-
collate information on methods used (e.g. preparation, instrumental technique);
-
operate regular cycles or rounds and provide accessible reports;
-
generate commercial quality control samples;
-
provide material to be used for learning, e.g. in technical workshops
Reference Materials
-
require the use of definitive analytical methods to derive reference values
-
homogeneity, long-term stability, continuous availability
-
can generate certified values for additional constituents over time
-
may be solutions (e.g. smoke condensate) as well as tobacco products
25
Thank you for your attention
Questions?
www.bat-science.com
4. Mainstream Cigarette Smoke Constituents
4 regional markets were sampled and representative products tested
• Market A
– Mixed Blend, low incidence of charcoal in filter, Sampled Q1 2012
• Market B
– Mixed Blend, High incidence of charcoal in filter,
Sampled Q4 2012
• Market C
– Predominantly FCV, Sampled Q1 2013
• Market D
– Predominantly US-Blended, Sampled Q4 2013, largest market
Thanks to my colleague Alison Eldridge of BAT Group Research & Development.
The data and a more complete discussion of their implications will be published
during 2014 and beyond.
27
4. Summary Statistics (4 markets combined)
Statistics for all replicates.
Products tested under HCI conditions
Market
Toxicant
Units
N
Mean
StdDev
Min
P25
Median
P75
Max
A
1,3-Butadiene
µg/cig
925
106.9
14.2
63.3
97.4
107.3
116.4
147.3
A
B[a]P
ng/cig
925
18.0
4.6
8.2
14.8
17.7
20.7
43.2
A
NNN
ng/cig
925
68.0
56.8
0.0
31.2
51.2
79.4
376.5
B
1,3-Butadiene
µg/cig
695
70.8
16.2
31.4
60.3
68.1
79.3
126.3
B
B[a]P
ng/cig
700
18.6
4.9
7.1
15.2
18.1
21.3
42.2
B
NNN
ng/cig
700
139.1
52.4
31.0
104.0
135.5
163.0
423.0
C
1,3-Butadiene
µg/cig
840
97.1
15.9
50.0
88.4
99.1
108.0
140.9
C
B[a]P
ng/cig
840
20.2
5.4
10.4
16.2
19.4
23.7
38.5
C
NNN
ng/cig
840
142.5
69.5
12.6
101.5
132.8
173.0
477.7
D
1,3-Butadiene
µg/cig
700
98.8
20.5
51.8
83.1
101.3
112.9
156.5
D
B[a]P
ng/cig
700
13.1
3.9
5.1
10.4
12.7
15.3
30.3
D
NNN
ng/cig
700
150.4
78.1
33.2
102.9
133.4
174.4
515.2
Total number of Products tested = 3165, replicates of 5
28