Passive smoking under controlled conditions

Intnaflonal Ardhlv of

Int Arch Occup Environ Health ( 1985) 56:99-110 40 tlia tiainl

© Springer-Verlag 1985

Passive smoking under controlled conditions

L C Johnsonl, H Letzel l , and J Kleinschmidt 2

1 Gesellschaft fiir Informationsverarbeitung und Statistik in der Medizin e V ,Studie Passivrauchen und Lungenkrebs Entwicklungsphase, Pettenkoferstra Be 35,D-8000 Miinchen 2, Federal Republic of Germany

2 Institut fiir Medizinische Balneologie und Klimatologie, Klinikum Gro 13 hadern,Marchioninistra Be 15, D-8000 Miinchen 70, Federal Republic of Germany

Summary Ten healthy subjects were exposed to passive smoking at a highlevel corresponding to 25-30 ppm CO in the ambient air for 3 h All subjectswere exposed at the same time in a climatic chamber especially designed forexposure experiments Despite an identical exposure rate considerableinterindividual variability of subsequent nicotine and cotinine levels insaliva, plasma and 24-h urine were observed This variability was more pro-minent in nicotine than in cotinine levels The kinetic pattern as reflected bysaliva levels for up to 24 h was consistent with previous data found in activesmokers Nicotine levels found in saliva were markedly influenced byrepeated sampling This was not the case for cotinine levels With regard tolaboratory techniques RIA seems to be more sensitive than gaschromato-graphy (GC) The results of this study suggest that measuring cotinine levelsin 24-h urine with RIA is presently the most sensitive and reliable criterionfor estimating exposure to passive smoking and for validating questionnairesor interviews about short-term exposure to passive smoking.

Key words: Identical exposure Nicotine Cotinine Saliva Urine -RIA Gaschromatography

Introduction

A number of studies have been published dealing with the relationship (or lackof one) between passive smoking and increased risk of disease, for example pas-sive smoking and lung cancer l 1, 4, 11, 14, 16, 21 l, respiratory problems inchildren (e g l 2, 5, 19 l), as well as the changes in lung function in adults (e g.l 3, 15, 231).

One of the thorniest problems in these studies is that of accurately assessingthe exposure to passive smoking.

Offprint requests to: L C Johnson at the above address

L C Johnson et al.

In order to clarify some of the current issues in this field an experiment wasdesigned to answer the following questions:

1 What is the kinetic pattern of nicotine and cotinine as measured in salivafollowing controlled exposure to passive smoking?

2 What is the variability within and between individuals after an identicaland high exposure to passive smoking ( 25-30 ppm CO for 3 h)?

3 Does the collection procedure for saliva influence the results?4 To what extent are the results of GC and of RIA the same and which of

these two methods commonly used is more appropriate for determining nico-tine and cotinine levels at low concentrations?

5 What is presently the most appropriate parameter for an objective assess-ment of exposure to passive smoking and for validating an interview aboutexposure to passive smoking during the preceding 24 h?

Methods

Subjects

Ten healthy volunteers between the ages of 18 and 65 ( 5 men and 5 women) took part in theexperiment They were all non-smokers who did not have a smoker in their household.

Exposure

A single exposure session of 3 h was conducted in a 60 m 3 climatic chamber especially design-ed for exposure experiments The conditions of exposure were controlled so that a high andconstant level of tobacco smoke was present in the room.

A smoking machine was used to generate the smoke-polluted atmosphere It was set so asto draw 15 S followed by a 15 S pause The mainstream smoke produced during the drawingphase of the cycle ( 89 ml of smoke per second) was routed outside of the climatized room sothat only side-stream smoke polluted the room atmosphere A total of 24 packs of cigaretteswere consumed during the 3-h exposure period.

A continuous reading CO measuring device (Comovarn 100 C, produced by Draeger,Libeck, FRG) was used to monitor the level of air pollution in the room A level of CO rang-ing between 25 and 30 ppm was maintained for the majority of the 3-h exposure interval

9 930 10 10

3011 1130 12 1230

Time of day

Fig 1 CO and nicotineconcentrations in the airduring exposure interval

CO l ppm

40

30

20

10

0

100


(Fig 1) As a check for the validity of CO as a tracer substance for the level of pollution, airin the room was sampled at 3 points in the time A measured volume (at least 101) was drawnthrough a tube containing a filter The mass of captured nicotine was subsequently determinedusing gaschromatographic techniques The concentration of nicotine obtained from thesesamples is also indicated in Fig 1 and shows that a high and relatively stable level of nicotinepollution was reached and maintained in the room during the experiment Ventilation was setat 20 % fresh air per time unit Due to the high concentration of side-stream smoke all subjectshad to wear goggles throughout the exposure phase of the experiment in order to avoid drop-outs because of unbearable eye irritation.

In order to assure that all individuals were presented with the same level of pollution theywere all exposed simultaneously in a single session in the same climatic chamber In addition,sitting positions were randomly changed at half hourly intervals.

Sample collection

Special attention was paid to the saliva collection procedure Drinking straws were used to-gether with especially cleaned plastic test tubes and stoppers Collecting through straws pro-ved to be convenient for the subjects and had the added benefit of limiting the contaminationof the sample from nicotine that might be present on the lips All subjects were advised to lettheir saliva freely flow through the straw into the test tube until a preset mark (equivalent to3 ml) was reached All specimens were deep frozen at -70 centigrade immediately after thecollection.

Saliva samples were collected immediately prior to and after exposure and thereafter atthree hourly intervals for 12 h A last sample was taken on the next morning, 21 h after theend of the exposure interval Twenty-four-hour urine was collected starting immediatelybefore the exposure interval.

Immediately after the end of the exposure interval a sub-experiment was undertaken inwhich the impact of the saliva collection procedure on measured concentrations of nicotineand cotinine was systematically manipulated Nine subjects were randomly allocated to threegroups Group I (controls) gave three saliva samples in rapid succession without stimulation.The other two groups gave a preliminary sample in the normal way but their second and thirdsamples were produced under two stimulated conditions: rinsing their mouths with water andlemon juice in a reversed order (Group II: water then lemon, Group III: lemon then water).

Laboratory techniques

Nicotine and cotinine were quantified by radio immuno assay l 17 l Selected specimens werereanalyzed by gaschromatography l 7 l so that a comparison between the two analytical meth-ods could be conducted.

Results

1 Saliva concentrations

At the beginning of the experiment low though measurable baseline levels wereobserved for nicotine and cotinine in a majority of the subjects (Figs 2, 3) Atthe end of the 3-h exposure interval high nicotine levels (median: 89 7 ng/ml;range: 247-186 5 ng/ml) were measured They were followed by a quick dropduring the next 3 h until almost baseline levels were reached again Cotinine,showing a less marked increase, reached its maximum 3 h after the end ofthe exposure (median: 19 5 ng/ml; range: 14-29 ng/ml) followed by a slowdecline All individual levels of nicotine and cotinine in saliva are also given inFigs 2 and 3 respectively Despite substantial variability in the nicotine values

101

102 L C Johnson et al.

Nicotine insaliva

Ecm 150-c

o

._

O 100-2C

0o

50

500-

c 100-

o50-

oas10-

1 -

8

1-

0,5-

Not detected -

Fig 2 Individual and pooled levels of nicotine in saliva (below: observed values; above:median, 25 % and 75 % quartiles, range)

I

MMMMMI EXPO �t

103Passive smoking under controlled conditions

Cotinine insaliva

E

° 15-C

25-0

()

0 3 6 9 12 15 2 h

50-

40-Ecm 30-

a_ 20-

C

c0 l10-C

, 5-Cc

o

1-Not detected - 3.

Fig 3 Individual and pooled levels of cotinine in saliva (below: observed values, above:median, 25 % and 75 % quartiles, range)

the overall trend is consistent with a short half life ( 30 to 110 min) Similarly thecotinine values are consistent with a substantially longer half life ( 19 to 40 h).The observed variability in the two substances is different Whereas nicotinemaintains a variability of roughly two orders of magnitude throughout theobservational period cotinine values were found to vary substantially less(roughly one half an order of magnitude).

r MMMMM

u

L C Johnson et al.

2 Comparison of analytical methods

For comparing laboratory methods matched samples of saliva, plasma and 24-hurine were analyzed by RIA and gaschromatography (Table 1) The time ofsampling was selected in a way that low as well as high concentrations could becompared between the two methods.

Table 1 Comparison of laboratory methods (RIA vs GC) Saliva and plasma levels are givenin ng/ml, urine levels in tg/24 h Only those samples are included which had detectable levelsby both methods Pre-exposure levels for cotinine in saliva are omitted because only onesample fulfilled this criterion

Sub Source Time of N x + S D p rstance collection RIA (sign(hours) RIA Gaschromato-

(hours) graphy test)

Nicotine Saliva 0 8 3 81 + 4 06 3 87 ±+ 1 89 1 00 0 78(ng/ml) 6 10 30 3 + 40 57 37 6 + 52 95 0 51 0 85Plasma 3 5 3 54 + 1 58 5 00 + 1 68 0 37 0 22(ng/ml)Urine Continu 10 134 41 + 95 84 105 3 + 88 4 0 04 0 93(ng/ml) ously

( 24 h)

Cotinine Saliva 6 8 19 7 4 58 18 72 ± 6 25 1 00 -0 43(ng/ml)Plasma 3 9 9 00 ± 4 88 8 88 + 5 31 0 73 0 71(ng/ml)Urine Continu 10 110 21 + 54 75 82 83 + 47 3 0 02 0 93(ng/ml) ously

( 24 h)

Interindividual variance is considerable However, neither saliva nor plasmalevels show statistically significant differences between the two laboratories:The correlation between the laboratories varies between r = 0 22 (plasma nico-tine) and r = 0 93 (nicotine in 24-h urine).

The analysis of the 24-h urine data showed different results The meansfound by RIA are approximately 30 % higher than those of gaschromatography(p < 0 05) This holds for nicotine as well as for cotinine There is a high correla-tion between the two methods for both variables (r = O 9 3 ,p< O 05).

In order to compare the sensitivity of the two methods, values for saliva andplasma levels were classified as detectable or not detectable (Table 2) Forplasma, the samples collected immediately before and after exposure werepooled Despite the small number of observations there is a consistent patternindicating that RIA may be more sensitive than GC Altogether there were 18samples with levels detectable by RIA but not detectable by GC The oppositewas found for two samples only The high incidence of samples without concen-trations detectable by either method was not surprising since two thirds of thesesamples represent baseline levels.

104


Table 2 Comparison of sensitivity between RIA and GC Saliva: baseline levels Plasma:pooled results from samples collected immediately before and after exposure

Saliva PlasmaDetectable by GC Detectable by GC

no yes no yes

Nicotine Detectable no O 1 10 0by RIA yes 1 8 5 5

Cotinine Detectable no 3 0 4 1by RIA yes 6 1 6 9

3 Influence of saliva collection procedures

The design randomized blocks repeated measurements is illustrated in Fig 4.Basically, there are three questions that can be investigated by analyzing thiscomplex design:

1 Are the groups comparable, i e has the randomization procedure work-ed? This is answered by comparing sample 1 between the groups.

2 Does repeated sampling itself influence the nicotine and cotinine levelsfound in saliva after exposure to passive smoking? For this questions, all threesamples are compared within group I.

3 How is the concentration of the two compounds affected by saliva secre-tion stimulation?

(n= 3)

Group (n 3)

III(n:3)

Sample

II

nostimulation

lemon -juice

water

III

nostimulation

water

lemon-juice

I

nostimulation

nostimulation

nostimulation

Fig 4 Experimental design for investigating the influence of collecting procedures on salivalevels

Table 3 summarizes the results Comparing sample 1 between the threegroups, very little variation can be discovered between the groups with regardto cotinine (ANOVA: p = 0 58), but there appears to be some inhomogeneityregarding nicotine: the concentration of groups I and III is twice as high as thatof the one in group II ANOVA results (p = 0 16) suggest, however, that thisdifference may be accidental.

Group I produced three saliva samples without additional stimulation Theaverage observed levels for these samples vary substantially for nicotine Thedramatic drop (approximately 50 % between sample 1 and 2) is statistically

105

L C Johnson et al.

Table 3 Nicotine and cotinine levels (ng/ml) in saliva under stimulated and unstimulatedsecretion conditions

Sample

1 2 3

Nicotine Group I 98 0 ± 24 6 59 5 + 41 5 48 1 ± 33 0Group II 52 4 + 37 0 17 9 ± 13 6 26 2 + 24 3Group III 106 0 ± 32 4 43 3 ± 28 4 16 9 ± 5 1

Cotinine Group I 16 7 ± 1 5 15 7 ±+ 6 1 21 7 + 4 0Group II 15 3 + 3 8 14 0 + 1 0 15 3 + 2 5Group III 18 3 ± 5 0 17 7 ± 5 7 15 3 + 2 5

significant (ANOVA: p < O 05) This was not the case for cotinine (ANOVA:p = O 13).

For answering the third question the differences in concentration betweenthe first two samples were compared between the three groups These differ-ences were used not only to adjust for potential inhomogeneities of sample 1between the groups, but also to remove the repeated sampling effect demon-strated above ANOVA results (nicotine: p = 0 29 ; cotinine: p = 0 98) suggestthat stimulating the secretion of saliva does not noticeably influence the concen-tration of nicotine or cotinine in a second sample.

Discussion

A number of studies have been published which approach the problem of as-sessing exposure to passive inhalation in one form or another An early study byFeyerabend et al l 8 l is typical in its design and can serve as a model for most ofthe studies that have been published to date Using questionnaire techniquesnon-smokers were classified as to their exposure to tobacco smoke through pas-sive smoking Subsequently, a particular body fluid (in some studies: expiredair) was sampled and analyzed with regard to a particular objective tracer sub-stance Of interest is the extent to which the exposure classification effectivelyexplains the variation observed between individuals.

In Table 4 we summarize a review of seven such studies These studies areheterogeneous with regard to the population examined (children, adults, onlywomen), the tracer substance (CO, CO Hb, nicotine, cotinine, thiocyanate,etc ), body fluid (expired air, urine, blood, and saliva), and laboratory tech-nique One thing, however, which all of these studies have in common is a largerange of observed values (from one to two orders of magnitude depending uponstudy and tracer substance) In trying to determine the effectiveness of a classi-fication instrument this variation leads to interpretation difficulties This stemsfrom the fact that subjective and objective assessment of exposure are com-pletely confounded with actual exposure differences as well as individual kineticdifferences between individuals Hence, when one observes a substantial

106


0 ) . b~ ,

3: "O-DC o= o C

=.Ooo C O O

Z" e R 5 D B ,,

-1 Ci 1 r'i r

Ct C )C0

coO 3 u

c c m

o 0.

, rn "

'.= 0.

,-'Z c o F

d~ -.C) C

o ,

Ct)) O C)C) C O 0-

l C

'C

.P

UUX C)

* O

C E

. U8 at O Wr Z

'ZV

Cli

CC

0CC)

.2 U 2 -0 a F c, t8 =

z 8 X UU~,

,UW O

0 , = 00

O 00 C )

C O 8 O

x O Z O > 0: =Q(U O Z p :5 v,'

C

0

W O Q

C Ow

.o

0 o 0 >

C

-E0000 >

or O 0

O g

0

'c 3

(U

o =

0

2 E

0 x*-0 2 ~EU .M 00,E M

0

EC)0

0

c

c

0 ~~~~~~0 ) O ~c~.2 02 C

0 o 04 u O C 1 O0 a C t

C) O g E 4 o E o

CZ VE d W 1

O 2 2 2 CCC)2 ·

U~ 3 O = 2 o 2 =LC~ O 4:,CC)

CC O C C C O (

.C " "eo C C~ U C):~C) 4 00 C

107

C)V o0 C

Oc

.iz O0 3

o Ct

C r

.-C)

f: Cd C

.o

ac

C

U.2

b -00 -o

0o

.M.0 -)

C0

00.x

0

0

C

E

e11

cC).0

r)

U

E

Q>D0

> O_ )C

ZC04.0

Z

C;

0 Ct O

0 O

0E

0 >

0 >

Cr.

c C

0Ce'O t

C

,o

C O O '

V CZ c

00Z

.C

c

.200

0

0C)

0

. r.40 >

0O Ut

Oc00

z 1

ed O -

..C _a Oc

O

oK00.0.

0

C)

C)

0

0.0.

C)3r-

0

0

C)

14.2.0

4

Xe< ? 20 U

e 4 E

L C Johnson et al.

variance in observed objective measures it is not clear to what extent this hasresulted from true differences in exposure, interindividual kinetic differences,and outright misclassification errors.

The purpose of the experiment reported here was to determine the variabil-ity of nicotine and cotinine levels within and between individuals Great effortwas taken to make sure that all subjects had an identical level of exposure.Studies of this type are a necessary step towards accurately assessing exposurein epidemiological studies (e g with regard to investigating the relationship be-tween passive smoking and lung cancer) In case-control studies the exposure topassive inhalation must be estimated for (life-)long periods in the past This isonly possible by using questionnaires or interviews There is a need for validat-ing such a questionnaire before trusting the data We propose the followingprocedure:

1 Using equally structured questionnaires, scorable data can be collectedfrom a person regarding the preceding 24 h (in hourly intervals) as well as thetotal previous life (in yearly intervals).

2 Selecting an appropriate indicator substance, a correlation between theresults of the 24-h interview and levels of indicator substance found in bodyfluids (saliva, plasma, urine) can be determined If it is high, the method isvalidated.

3 Having validated the 24-h period it is a matter of believing in analogieswhether or not one also trusts a similarly structured life-long history.

In that context, the experiment reported in this paper aims at an approachto validating the results of short-term questionnaires about passive smoking sothat future epidemiological studies can be planned more efficiently The resultsshow that nicotine and cotinine levels are best measured using RIA The supe-rior sensitivity of RIA is essential since passive smoking under real life condi-tions leads to very low levels in body fluids.

Combining the work of other investigators with that reported here leads tothe following summary with regard to the most appropriate tracer substance.Levels of expired carbon monoxide and/or carboxyhemoglobin in blood are notappropriate since a variety of alternative sources for CO exist (lack of specific-ity) Similarly, thiocyanate has been shown to be inferior to cotinine bothbecause it is influenced by diet and because it is relatively insensitive at lowexposure levels l 13 l Nicotine and cotinine can both be measured in blood,urine or saliva Blood and urine are substantially more difficult to obtain underfield conditions than is saliva In addition, as this study shows, blood levels arevery low even after heavy exposure Both urine and saliva have the potentialdrawback of being situationally dependent Nicotine excretion by urine may berelated to its p H l 6 l Saliva concentrations of nicotine can be influenced by theconditions under which the sampling is conducted (repeated sampling) For allof these reasons nicotine's use as a tracer substance is inadvisable Cotinine, onthe other hand, appears to be well suited for the task.

The body fluid of choice seems to be urine Although measurable levels ofcotinine were observed in this experiment both in saliva and urine, it must beconsidered that our experiment was carried out under extreme exposure condi-tions which will hardly ever be reached in real life situations Our results make

108


it seem unlikely that cotinine levels measured in saliva are sensitive enough forvalidating purposes under real life conditions The cotinine levels obtained from24 h urine suggest that an exposure at one tenth of the level used in this experi-ment would presumably still lead to detectable levels in 24-h urine Of course,urine samples collected for a shorter period of time also contain measurablecotinine concentrations which, when related to urinary creatinine concentra-tion, can yield meaningful results.

The methodological tools are now available to validate interviews or question-naires on short-term exposure Such a validation stategy as outlined in thispaper could lead to methods of assessing the exposure to passive smoking whichincrease the chances for valid epidemiological studies on potential hazards ofpassive smoking.

Acknowledgements We wish to thank Prof Dr med K Uberla (Institut fir MedizinischeInformationsverarbeitung, Statistik und Biomathematik der Ludwig-Maximilian-Universitat,Miinchen) for his valuable methodological advice We are indebted to Dr Nancy Haley,American Health Foundation, Naylor Dana Institute, Valhalla, NY 10595, and her colleagueswho gave valuable methodological advice.

References

1 Chan WC ( 1982) Zahlen aus Hongkong Miinch Med Wochenschr 124:162 Colley JRT, Holland WW, Corkhill RT ( 1974) Influence of passive smoking and parental

phlegma on pneumonia and bronchitis in early childhood Lancet 2, 7888:1031-10343 Comstock GW, Meyer MB, Helsing KJ, Tockman MS ( 1981) Respiratory effects of

household exposure to tobacco smoke and gas cooking Am Rev Resp Dis 124:143-1484 Correa P, Pickle LW, Fontham E, Lin Y, Haenszel W ( 1983) Passive smoking and lung

cancer Lancet 2, 8350:595-5975 Ferguson DM, Harwood LT, Shannon FT, Taylor B ( 1981) Parental smoking and lower

respiratory illness in the first three years of life J Epidemiol Comm Health 35: 180-1846 Feyerabend C, Russell MAH ( 1978) Effect of urinary p H and nicotine excretion rate on

plasma nicotine during cigarette smoking and chewing nicotine gum Br J Clin Pharm5:293-297

7 Feyerabend C, Russell MAH ( 1980) Assay of nicotine in biological materials: sources ofcontamination J Pharm Pharmacol 32:178-181

8 Feyerabend C, Higenbottam T, Russell MAH ( 1982) Nicotine concentrations in urine andsaliva of smokers and non-smokers Br Med J 1: 1002-1004

9 Foliart D, Benowitz NL, Becker CE ( 1983) Passive absorption of nicotine in airline flightattendants N Engl J Med 308:1105

10 Friedman GD, Pettiti D, Bawol RD ( 1983) Prevalence and correlates of passive smoking.AJPH 73:401-405

11 Garfinkel L ( 1981) Time trends in lung cancer mortality among non-smokers and a noteon passive smoking J Natl Cancer Inst 66:1061-1066

12 Greenberg RA, Haley NJ, Etzel RA, Loda FA ( 1984) Measuring the exposure of infantsto tobacco smoke N Engl J Med 310:1075-1078

13 Haley NJ, Axelrad CM, Tilton KA ( 1983) Validation of self-reported smoking behavior:Biochemical analyses of cotinine and thiocyanate AJPH 73:1204-1207

14 Hirayama T ( 1981) Non-smoking wives of heavy smokers have a higher risk of lungcancer: a study from Japan Br Med J 282:183-185

15 Kauffmann F, Tessier JF, Oriol P ( 1983) Adult passive smoking in the home environment:a risk factor for chronic airflow limitation Am J Epidemiol 117:269-280

16 Knoth A, Bohn H, Schmidt F ( 1983) Passive smoking as cause of lung cancer in femalesmokers Med Klinik 78: 54-59

109

110 L C Johnson et al.

17 Langone JJ, Gjika HB, Van Vunakis H ( 1973) Nicotine and its metabolites: radioimmuno-assays for nicotine and cotinine Biochemistry 12:5025-5030

18 Matsukura S, Taminato T, Kitano N, Seino Y, Hamada H, Uchihashi M, Nakajima H,Hirata Y ( 1984) Effects of environmental tobacco smoke on urinary cotinine excretion innon-smokers Evidence for passive smoking N Engl J Med 311: 828-832

19 Rantakallio P ( 1978) Relationship of maternal smoking to morbidity and mortality of thechild up to the age of five Acta Paediatr Scand 67:621-631

20 Russell MAH, Cole PV, Brown E ( 1973) Absorption by non-smokers of carbon monoxidefrom room air polluted by tobacco smoke Lancet 1, 7803:576

21 Trichopoulos D, Kalandidi A, Sparros L, Mac Mahon B ( 1981) Lung cancer and passivesmoking Int J Cancer 27: 1-4

22 Wald NJ, Boreham J, Bailey A, Ritchie C, Haddow JE, Knight G ( 1984) Urinary cotinineas marker of breathing other people's tobacco smoke Lancet 1, 8370: 230-231

23 White JR, Froeb HF ( 1980) Small airways dysfunction in nonsmokers chronically exposedto tobacco smoke N Engl J Med 302:720-723

Received October 16, 1984 / Accepted March 3, 1985

Documents

Passive smoking under controlled conditions