Tobacco Documents Online

Prtrr• r.: c. Rrnrhe.mrcrn 3 Brha.:nr '~oi i7. pp. 17-:5 '- Peramon Press pl:. 1191 Pnnted m tne L'.S y r~tui-?Us"90 S-~ CXI , I I I I I I I I I I I I I I I Population Characteristics and Cigarette Yield as Determinants of Smoke Exposure RAYMOND B. BRIDGES. ` JUDY G. COMBS Department of Oral Health Science, College of Dentistry, Uttiversin• of Kentuckt• JANICE W. HUMBLE. JOHN A. TURBEK Tobacco and Health Research Institute, Universin• of Kenruck-Y STANLEY R. REHM Department of Medicine, Universirt• of Ketttuckt. L,e.rington• KY AND NANCY J. HALEY American Health Foundation, Valhalla. NY Received 20 July 1989 BRIDGES. R. B.. J. G. COMBS. J. W. HU?`4BLE. J. A. TURBEK. S. R. REHM AND N. 3. HALEY. Population characteristics and cigarette vteld as determinants of smoke exposure. PHAFLMACOL BIOCHEM BEHAV 37(1) 17-28. 1990.-Relationships of population charactencucs. smoh-tne historv. and cigarette yield with smoke exposure as measured by peripheral blood concentrations of thiocvanate. carboxvhemo¢lobin. nicotine and counine were soueht in 170 male smokers. This population of smokers had stgnificant elevations of serum thtocvanate. blood carboxvhemoglobin and plasma nicotine and counine concentrations as compared wtth an equal number of age- and sex-matched nonsmokers and these concentrations correiated significantly with past 24-hour ctzarette consumptton. Although the nicotine yield of the cigarette correiated significantly with plasma cotinine and marginally with plasma rucoune. the reduction in plasma nicotine and cotinine was not proportionate to the reduced yield of the cigarettes. suggesting that smokers partially compensate for the lower yields of their cigarettes. Blood levels of carboxvhemoglobin. nicotine and cotinine were also significandv associated with the weight of the subjects, presumably due to the relationship between weight and the volume of distribution. Univanate and multiple regression analyses provided evidence that coffee and alcohol consumption and years smoked also may be important determinants of smoke exposure. ~ ~ ~ CIL; ~ ~ Thiocvanate Carboxyhemoglobin Plasma nicotine Plasma cotinine Cigarette yield Body weight ~ Coffee Alcohol Smoking history CIGARE'I'I'E smoking is associated with an increased incidence of both respiratory and cardiovascular disease as well as cancer (49-51); however, many life-lono smokers suffer no such impair- ment of health. It is likely that both the dose of smoke constituents and the individual response of smokers to these constituents account for the variable susceptibility of smokers to these diseases. In an attempt to reduce the intake of tar and nicotine. commercial cigarettes have been developed which produce lower yields of these components under standardized smoking conditions (48). Although lower yield cigarettes are associated with lower plasma nicotine concentrations (19.45). no consistent relationship has been observed between the nicotine yield of the cigarette and plasma cotinine concentrations (4). This lack of a relationship suggests that the smokers of low yield cigarettes may compensate 'Requests for reprints should be addressed to Ravmond B. Bridges. Ph.D.. Department of Oral Health Science. College of Dentistry, University of ' Kentucky, Lexington, KY 40536-0084. 17

I I8 BRIDGES ET AL. I I I I I I 1 I I I I I I I I I I in their puffing behavior to derive a greater nicotine deliverv than that obtained under machine smoking conditions. Increases in puff volume, puff duration and puff number have all been associated with decreased nicotine yield of the cizarette smoked (2. 15. 37). These compensatory changes in puffing behavior may obliterate any health benefits which might accrue from smoking low tar and nicotine ciearettes. Nicotine has been suggested to be the primary pharmacologic reinforcer controlling puffing behavior (39). Previous studies have demonstrated that blood concentrations of smoke constituents and their metabolites are highly variable among smokers. In the context of a larger study to determine the relationship between cigarette smoking and the development of obstructive pulmonary disease, we sought the major determinants of smoke exposure and absorption as measured by blood levels of thiocvanate. carboxyhemoeiobin. nicotine and cotinine in 170 smokers. The purpose of the present study was to determine how population characteristics, smoking history (cigarettes per day, years smoked, and pack-years). and nicotine yield of the ciearette related to smoke exposure and absorption as determined by blood levels of these smoke constituents and their metabolites. TABLE I POPULATION CHARACTERISTICS' Nonsmokers Smokers P Age (years) 37.1 = 0.8 37.8 = 0.8 0.52 Height (m) 1.792 - 0.006 1.788 = 0.006 0.65 we,ght(kg) 79.7 = 1.0 77.6 x 0.9 0.12 Body Mass Index 24.8 - 0.3 24.3 = 0.3 0.19 Tea Consumption (cupsiday) 1.01 = 0.12 0.94 = 0.1~ 0.70 Coffee Consumption 1.69 = 0.18 3.84 = 0.30 0.0001 (cups/day) Alcohol Consumption= Present (ounces/wk) 2.57 x 0.29 8.73 = 1.06 0.0001 Cumulative (ounces) 2249 = 316 10254 = 1341 0.0001 'Results are expressed as mean = S.E.M. for 161 smokers and 168 nonsmokers. TAlcohol consumption is expressed as equivalent ounces of pure alcohol consumed. .%MTHOD Male smoking and nonsmoking volunteers for this study consisted of faculty, staff and students of the Universitv of Kentucky as well as individuals recruited by advertising in the local newspaper. All subjects gave written informed consent for this study, which was approved by the Human Investigations Committee of the University. Subjects were excluded from this study if they were taking medications or had diseases known to affect inflammatory mediators or if thev had a history of broncho- spasm or asthma. Male smokers were recruited without regard to their age, the brand of cigarette smoked. the number of cigarettes smoked per day or the pack-years smoking history. The 170 smokers were compared with a group of 170 age-, sex- and race-matched nonsmokers who had never smoked on a regular basis. The subjects for this study completed a detailed question- naire concerning medical. pulmonary, smoking (both active and passive) and drug usage histories. In addition. these subjects recorded their alcohol and coffee consumption. exposure to environmental pollutants, and demographics. Smokers provided data on the smoking history questionnaire concerning the cigarette brand and number of cigarettes smoked per day for each five-year interval of their smoking history. The subjects also completed a questionnaire indicating cigarette con- sumption in the previous 24 hours. Smokers and nonsmokers who used tobacco in any other form (i.e., pipe, cigars or smokeless tobacco) were excluded from the data analyses. The smokers in this study consumed their own brand of cigarette. The nicotine vield of the cigarette was derived from the Federal Trade Com- mission using standard machine smoking conditions (21). Subjects for this study also provided data on beverage (alcohol. coffee. and tea) consumption for each five-year interval of their drinking history. For each of these five-year periods, the subjects indicated the weekly alcohol consumption as number of 12 ounce bottles or cans of beer, the number of 5 ounce glasses or 25.4 ounce bottles of wine, and the number of 2 ounce shots. 16 ounce pints. or 25.4 ounce fifths of distilled spirits. Alcohol consump- tion was converted to pure ounces of alcohol assuming a 3.8% alcohol content for beer. 10% for wine and 45% for distilled spirits. The total cumulative and dailv alcohol consumption were calculated from these data and expressed as total ounces of pure ethanol consumed during a lifetime or on a daily basis. respec- tively. Coffee and tea consumption were expressed as number of cups consumed daily. Smokers were requested to smoke ad lib and to smoke a cigarette 5 minutes prior to venipuncture. Venous blood samples were collected at 8 a.m. after an overnight fast. Blood carboxy- hemoglobin (expressed as 9c of saturation) was determined spec- trophotomethcally using a fresh whole blood sample and a CO-oximeter (Instrument Laboratories. Model 182) (30). Frozen (-80°C) blood samples were used to analyze for thiocyanate and plasma nicotine and cotinine. Serum thiocyanate (expressed as µmoles/l) was determined spectrophotometrically as previously described (10). Plasma nicotine and cotinine concentrations were determined bv radioimmunoassay; the inter- and intraassav varia- tions are 6~'c with a sensitivity of 1 ng/ml for nicotine and cotinine (27,'9). Statistical analysis of the data utilized Student's f-test for unpaired data and Pearson's correlates for linear regression anal- ysis. Significant differences between groups were determined by analysis of variance (ANOVA) while predictors of blood levels of smoke constituents or their metabolites were determined using multiple regression analyses. These analyses were accomplished using appropriate SAS (Statistical Analysis System. SAS Institute. Inc.. Cary, NC) programs and an IBM 3083 computer. RESULTS Characteristics of Populations A comparison of the smoking and nonsmoking populations ~ according to age, height, weight. body mass (Quetelet) index {i.e.. weight (kg)/[height (m)J'} (28), and consumption of tea, coffee n and alcohol is given in Table 1. Smokers consumed significantly (p<0.0001) more coffee and alcohol (both present and cumula- G7 tive) than did their nonsmoking counterparts. ~ CA ~ Smoking History The smoking history variables are given in Table 2. As indicated by the range of values, there was considerable variation in smoking history within this population. The normal mean cigarette consumption per day was indicated by the smoker in the questionnaire concerning the five-year blocks of smoking history while the past 24-hour cigarette consumption was indicated by the smoker for the number of cigarettes smoked in the 24-hour time period immediately prior to participation in this study. The number r © C~

I I 1 I I I I I I I I I E I I I I DE7FRM1tiANTS OF S?.9Ot:F. E7:POSURE TABLE : SMOKING HISTORY' Mean = S.E.M. ) Normal Cigarene 31.8 = 1.0 Consumption/Dav- Ciaarettes Smoked 24.8 = 1.0 Past 24 Hourst Years Smoked 20.3 - 0.9 Pack•Years 24.6 = 1.0 Nicotine Yield 0.98 = 0.02 of Cigarette- 19 TABLE -' BLOOD CONCE.NTRAT1OtiS OF SMOKE CO\STITL'Ei`7S' Nonsmokers Smokers` Ranize Serum Thiocvanate 98.3 = 2.6 161.2 = 3.7 8-80 (µ.moles/h (0-2-13.9) (46.9-390.4) Carboxyhemoglobin 2.2 = 0.1 7.4 = 0.2 5-70 (ric) (0-5.2) (2.8-15.0) Plasma Nicotine 2.3 = 0.3 31.1 = 1.3 3.7-53.7 (ngiml) , (0-2I.1) (2.9-94.4) 3.8--60.0 Plasma Counine 2.9 = 0.4 384.0 = 12.5 0.28-1.56 (ng ml) (0-39) (35-717) 'Results are expressed as means (- S.E.M. ) for 161 smokers. `Notmai daily cigarette consumption (t.e.. cagarenes per day) was significantlv (p50.0001) higher than the number of cigarettes smoked in the past 24 hours. =\icottne yield is expressed as mg nicotineicigarette as determined by the Federal Trade Corttmission i 21). Since 20 smokers consumed generic ct2arettes for which nicotine vield was not available, the mean nicotine vteld was calculated for 141 smokers. of cigarettes smoked in the past 24 hours was sisrtificantly (p<0.001) less than the normal mean number of cigarettes smoked per day possibly as a result of some of the smokers attempting to reduce their consumption. The mean nicotine yield of the ciga- renes smoked was 0.98 mg per cigarette. ranging from ultra-low yieid cigarettes (0.28 mg per cigarette) to nonfiltered cigarettes (1.40 to 1.56 mg per cigarettel. Relationships Bemeen Population Characteristics and Smoking Histon• The relationships between population characteristics and smok- ine history are siven in Table 3. Age was significantly related to parameters of cumulative smoking history (i.e.. pack-years and years smoked) and nicotine yield of the cigarette, but not daily cigarette consumption. Weisht. bodv mass index and tea con- sumption were not associated with any of the indices of smoking historv. Increased coffee and alcohol consumption were associated with increased normal daily cigarette consumption. cigarettes smoked in the past 24 hours. years smoked and pack-years "Results are expressed as means (= S.E.M. ) for 161 smokers and 168 nonsmokers. The ran¢e of concentrations is indicated in the parentheses. TConcentrations of all smoke constituents were significantly (ps0.0001) higher in smokers, smoking history. In addition. increased aee in smokers (but not nonsmokers) was significantly (p<0.0008) associated with coffee consumption (r= .31:) and cumulative (but not present) alcohol consumption (r= .263). Blood Conccntrarions of Smoke Constituents and Their Interrelationships As shown in Table 4, the smokers had significantly (p<0.0001) higher blood levels of thiocyanate. carboxyhemoglobin, nicotine and cotinine than the nonsmokers. There was considerable varia- tion in these concentrations with overlapping values being ob- served within the smoking and nonsmoking populations. Fourteen nonsmokers had a carboxyhemoglobin level greater than 3%, whereas only I smoker had a carboxyhemoglobin level less than 3~c (data not shown). Plasma nicotine and cotinine concentrations were more useful in differentiating smokers from nonsmokers. Only 4% (7 of 161) of the nonsmokers had a plasma nicotine which exceeded 10 ng/ml in contrast to 97% (155 of 161) of the smokers. This relatively high plasma nicotine concentration in a few nonsmokers was possibly due to environmental contamination, since these values were not paralleled by comparable increases in cotinine concen- trations. Only 2 of 168 nonsmokers had a plasma cotinine which TABLE 3 CORRELATES-POPULATION CHARACTERISTICS A.D SMOKItiG HISTORY' Normal Daily Ciearette Consumption Cigarenes Past 24 Hours Years Smoked Pack- Years Nicotine Yield Aee .047 .016 .922~ .516T .167t Weieht .064 -.057 -.030 .037 .014 Body Mass Index .094 -.014 -.035 .037 .062 Tea Consumption -.073 .037 -.019 -.043 .071 Coffee Consumption .30V .134 .3921 .427~- ?18t Alcohol Consumption Present .:87§ .233T .122 .217` .103 Cumu lative .3444 .213t .350T .4589 .208' 'Pearson's correlates for 161 smokers, except for nicotine yield (n= 141). 1ps0.05: +ps0.0l: §ps0.001: Sp:50.0001. I

1 ~ ~ ~ i 20 50 ~ BRIDGES ET.-1L. 90 !10 70 60 50 40 30 a 20 lo 0 r=CD519, •~1.80 = 0308 P ~ Q=1 0 50 1Do 150 200 250 300 350 400 a5C ~0 550 600 650 700 750 0 50 Y X 150 200 250 300 330 400 450 5C0 S50 600 PLASMA COTININECng/mD 650 700 750 6hB y s01x787a •4.1056 - 0376 iR '5 k ~ 14 7 =0.0001 . 13 Z ~ J 12 O d 11 V ~ 10 ~ W S 9 W r 8 Y 0 m 7 X m ~ Q U S ~ G 4 U O 3 O J (D 2 I A J 0 yz0.0792: •13 1.0 =0.270 p = 0.00 1 PLASMA COTININE (ng/ml) PLASMA NICOTINE Cng/mU FIG. 1. Linear reiationships among the blood concentrations of cigarene smoke constituents. The graphical representation for each of these relationshios includes the eouatlon for the insettee least-squares best fit line: the correlation coefficient (n: and the level of si¢nificance for the correlation. The data are given for the total population (n = 16 1) smoking all brands of ci¢arenes. exceeded 20 n_vml. while only 5 of 161 smokers had a plasma cotinine concentration of less than 100 ngiml. The possibility cannot be preciuded that some nonsmokers were "closet smok- ets." No relationship was obsetved between these plasma concen- trations of nicotine or cotinine in nonsmokers and their self- assessment of passive smoke exposure (data not shown). Nicotine. cotinine and carboxvhemo¢lobin correlated with one another in a highlv significant manner (r>.508. p<0.0001) as shown in Fig. 1B. C and D. Serum thiocyanate concentration also correlated signiftcantly with plasma cotinine (Fig. lA), but not with carboxyhemoglobin (r=.111, p= 0.14) or plasma nicotine (r=.113. p=0.161. Relationships or' Blood Concentrations of 5moke Constituents or Their Metaboiires to Population Characteristics The relationships between population characteristics and blood concentrations of smoke constituents are given in Table 5. Si¢nificant nezative correlations were observed between wei¢ht and the concentrations of carboxyhemoglobin. nicotine and coti- nine. Also. signiticant or marginally significant relationships were observed betAeen the body mass index and thiocyanate. nicotine and cotinine concentrations. Coffee and cumulative alcohol con- 0 50 00 150 200 250 300 350 400 450 500 550 600 650 700 750 PLASMA COTININE (ng/ml) r k° 9~ e 7 ~ 6~ Sr 4 r 3r 2! r:0.0916: •4.5432 •.0.623 0.0001 0 L C S: :5 20 25 30 35 40 45 50 55 60 65 70 75 00 85 90 95100 sumption were also significantly associated with plasma nicotinec_~ concentrations. ~ ~ TABLE 5 C,G ~ ~ CORRELATES-POPUL?.TION CHARACTERISTTCS AND BLOOD CONCENTRATIONS OF SMOKE CONSTITUENTS' ~ Cn Carboxy- Thiocyanate hemoglobin Nicotine Cotinine Age .070 .068 .116 .091 Weieht -.139 -.195+ -.190t -.1'9= Body Mass Incex -.15-1' -.107 -.153 -.163` Coffee Consumonon -.031 .066 .212= .13: Alcohol Consumption Present .031 .113 .073 .053 Cumulative .052 .128 .265§ .150 `Pearson s :;rrelates for 161 smokers. 'p:S0.0=: _-50.0L• :ps0.001.

DETFR`,11\A:r"TS OF S.NSpKE EtiPOSURE. 1 ~ ~ ~ 1 TABLE 6 CORREL.-NTES-SMOI:ING HISTORI'.-%.ND BLOOD CONCfi.NTRAT10NS OF SMOKE CONSTITUEN'I'S• Thtocvanate Carboxv- hemoglobin Nicotine Cotinine Normal Ctgarette Consumctton/Dav .119 .254= .'_45- .238, Cigarettes Smoked Past 24 Hours .1 .332§ .259r .372§ 1'ears Smoked .0€' .001 ?19e .121 Pack-Years .1'_0 .150 .3„§ .218; Nicotine Yield of Cigarette - .0:7 .014 '38` ?20+ "Pearson's (linear) correlates for 161 smokers. except for nicotine yield (n= 1.411 . -ps0.05: =p:S0.01: §ps0.0001. Relationships of Blood Concentrations of Smoke Constituents to Smoking HistonThe relationships between blood concentrations of smoke constituents and indices of smoking histon• are given in Table 6. Significant correlations between thiocvanate. carbox}•hemoglobin. 400 ~ 350 a ~ 300 00 A ..o. ~0 ,.0.02 . : . 50f!'. . _ _- . . . ~ 0 5 .0 i S 20 30 40 50 60 70 CIGARETTES PAST 24 MWRS 5 10 15 20 30 40 50 60 70 nicotine or cotinint concentrations and past 24-hour cigarette consumpuon were all observed. Although sicnificant correlates were also observed between these concentrations and the normal daily cigarette consumption. the past 24-hour cigarette consump- tion was alwavs the better predictor. Also. the loganthmic transformation was shown in each case to increase the prediction of the blood concentrations of these smoke constituents (rather than the linear plot) (Fig. 2). The log of the past 24-hour ctgarettes consumption was the best predictor of plasma cotinine concentra- tion (accounting for 13.9c'K of its vanation), followed in order bv carboxvhemoelobin (10.19c). nicotine (5.8CC ) and thiocvanate (3.2%). Increased plasma nicotine and cotinine concentrations were both significantlv associated with pack-years smoking his- torv and the nicotine yield of the cigarette, whereas oniv increased plasma nicotine concentrations were significantly associated with vears smoked. ~ Nicotine Yield The significant association between nicotine vield of the cigarette and age, ~ears smoked or pack-years smoking history suggested that the smokers consuming different yield cigarettes might also differ in population characteristics or indices of smoking history. The smoking population was. therefore. arbi- trarilv divided into six groups according to relativel~, narrow ranges of nicotine vield of their cigarette (Table 7). Comparing the characteristics of these groups. smokers consuming nonfilter '0 i eo . C z 0 U J 20 a 0 0' 900 eoo E . .0.24 1 2 ' r.0.00 5 10 $5 20 3C 40 50 60 70 .D oL, CIGARETTES PAST 24 HOURS r.0.3 7 3 Os 0.000 i : /. 5 to 15 20 30 40 50 60 70 CIGARETTES PAST 24 NOURS CIGARETTES PAST 24 HOURS FIG, 2. Relationships between blood concentrations of smoke constituents with cigarette consumption in the past 24 hours. The log-linear (cigarettes past 24 hours - concentntion of smoke constituent) graphical representation in each case provided a more highly significant correlation and least-squares best fit than did the linear relationship. The correlation coefficient trf and the level of cignificance are given for each of these relationships. The data given are for the total population (n a 161) smoking all brands of cigarettes.

I I I I I I I I I I I I I I I I I I „ TABLE 7 CHARACTERISTICS OF SMOKERS ACCORDING TO `ICOTINE YIELD OF THEIR CIGARETTE` Group Ranee of Nicotine Yield N pee (years) Cigarettes Past 24 Hours Years Smoked Pack- Years 1 0.28-0.43 5 36.1 = 2.5 19.4 = 5.4 16.8 - 3.0 =3.8 = 7.2 2 0.50-0.70 16 35.2 = 2.2 23.1 = 2.6 15.9 = 1.7 20.2 = 2.6 3 0.71-0.90 22 40.0 = 2.8 24.7 = 6.6 20.5 = 2.9 23.5 = 6.7 4 1.05-1.10 65 :6.1 = 1.2 25.6 = 1.6 19.1 = 1.2 24.1 - 1.5 5 1.11-1.20 17 30.6 - 1.4= 24 . 2= 2.6 13.5 = 1. 5' 16.4 - 2._'' 6 1.40-1.60 14 48.4 = _.1- 25.1 = 3.0 33.9 = 2.6' 37.9 = 3.2T 'Data are expressed as mean ( = S.E.M.) for the indicated number of subjects in each group. `• Groups 5 and 6 differed significantly (ps0.05) from the other 4 oroups in mean age. years smoked. and pack-years smoking history. Group 6 also differed from the remaining fi%e groups in having a higher consumption of both coffee and alcohol (data not shown). ciearettes (Group 6) had sigttificantly higher mean age, years smoked. and cumulative smoking history (pack-years), while smokers in Group 5 had significantly lower mean age. years smoked and pack-years smokin¢ history. In addition. smokers in Group 6 had a significantly higher coffee and alcohol consumption than did the other 5 groups (data not shown). The first » groups did not differ significantlv in population characteristics or smoking history charactenstics (normal cigarette consumption. cigarettes past 24 hours. pack-years smoking history or years smoked). These differences in the 6 smoking groups were consistent with the significant correlation observed between nicotine yield and pack- years smoking history (r=.196. p=0.02) or years smoked (r= .256. p=0.002). white nicotine yield did not correlate significantly with either of these parameters in the first 4 smoking groups. In order to avoid the complicating effects of the observed differences in population characteristics of Groups 5 and 6 vs. 1-4. further regression analyses were limited to Groups 1-4. The blood concentrations of smoke constituents in each of the six groups were also detetznined (Table 8). Smokers consuming the lowest nicotine vieldine ci¢arettes also had the lowest blood concentrations of thiocyanate, carboxyhemoglobin. nicotine and BRIDGES ET .-11.. cotinine. This effect was especially pronounced for smokers ir. Group 1. a.lthough the serum thiocyanate and blood carboxyhe- mo¢lobin concentrations were not si¢nificantlv different in the first four groups. plasma nicotine and cotinine concentrations appeared to increase progressively with increasing nicotine yield of the cigarette. These decreases in plasma nicotine and cotinine observed with decreasing nicotine yield of the cigarette were supported by the linear relationships between plasma nicotine or cotinine concen- trations and the cigarette yield in smokers consuming filter cigarettes (Groups 1--t) (Fig. 3). The nicotine yield of the cigarette smoked contributed marginally (p=0.08) to the prediction oi plasma nicotine (2.8%) and significantly (p=0.008) to the pre- diction of plasma cotinine (6.6%). However, the nicotine yield oi the cigarette did not correlate significantly with either serum thiocyanate concentration (r =.052. p= 0.40) or blood carbox~- hemoglobin level (r=.105. p=0.28). Carbon monoxide yield of the cigarette also did not correlate significantly with carboxyhe- moglobin level (r=.112, p=0.:0). .Although smokers of low yield cigarettes had lower plasma nicotine and cotinine concentrations than did smokers of higher TaBLE 8 BLOOD CONCENfR4TI0NS OF SMOKE CONSTITUE.N"rS L` GROUPS OF SMOKERS ACCORDING TO NICO'IINE YIELD OF TFEIR CIGARE7TE• ~ Meatt Nicotine Carbox,- - Plasma Plasma Yield Thiocyanate hemoglobin Nicotine Cotinine Group (mg)cigarette) (µmolesli) (%) (ng/ml) ng/ml) 1 S 0.34 132.2 = 17.5 5.7 = 1.0 18.0 = 1.4 :56=92 2 \~ 0.56 160.7 = 9.1 7.4 = 0.6 29.7 = 5.3 _:0 - 37 ~ ~ 0.76 175.9 = 11.3 7.6 = 0.5 38.4 = 3.1 351 - 28 ~ 4 bS 1.06 163.6 = 6.0 7.6 = 0.3 32.5 = 1.7 »09 = 19 1.16 143.8 = 8.9 7.9 = 0.7 29.8 = 3.6 332 = 45 ~ 6 ~ 1.-t8 162.3 = 10.2 7.4 = 0.7 44.4 = 5.7 459 = 42 Cr7 ~ % Chan¢e- -77% - 18.5q -2.1.0°c -59.4% -44.2ryc W 'Data are expressed as mean = S.E.M. with the number of subjects in each group given in Table 7. 'The percentage change in the means comparing Groups 6 and 1.

1 D-7TFRMItiANTS OF S11OKE EXPOSURE A r=0.0377:•6.579 ~6 F =C.206 5 I L e.0.0 3 5 4 r ~ i3 l 2 12 ~ 8 ~~ L 10I ~ 9 r ~ 1 1 ~ t 1 1 1 ~ r 1 1 1 1 1 1 900 0.1 0.2 0.3 0.4 r 0.5 0.6 0.7 0.8 0.9 1.0 NICOTINE YIELD (mg/cgaratt.) B r. ~ 74.0r4222.9 r:0.2 5 7 /•0.008 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 E \ P v z z r 00 0 ~,0. 1 0.2 0.3 NICOTINE YIELD (mg/cigaratts) FIG. 3. Relationships between plasma concentrations of nicotine or counine and nicotine yield of the cigarette. The graphical representation of these relationships includes the equation for the inserted least•squares best fi: line: the correlation coefficient: and the level of signiticance for the correlation. The data are etven for smokers smokmg filter cigarettes with mcoune ytelds of 0.28 to 1.10 mgrci¢arette (n= 108) d vteid cigarettes. these decreased concentrations were not propor- uonate to the decrease in yield of the cigarette smoked. For example• comparing smokers in Groups 6 and 1, there was a 77% decrease in the mean nicotine yield of the cigarettes with a corresponding 59.4% and 44.2% decrease in mean plasma nico- tine and counine concentrations, respectively. Similar results are obtained by comparing other groups (e.g.. Groups 4 and 1). This apparent compensauon for low yield cigarettes was not achieved by an increase in the number of cigarettes smoked per day. No significant difference in number of cigarettes smoked in the past 24 hours was observed in any of the six groups. Futther, there were no significant linear correlations between nicotine yield and normal daily cigarette consumption (r=.069• p=0.48) or cigarettes smoked in the past 24 hours (r=.110• p=0.26). Relationships of Blood Concentrations of Smoke Constituents to Cumulative Smoking Histore The relationships of blood concentrations of smoke constitu- enu with pack-years smoking history was also determined in the j 71~.~------. • . x 6F ¢ 5~ • r < 4 r U • 3 2 0 C 5 10 15 20 25 30 35 40 45 50 55 60 PACK-YEARS SMCKING HISTCRY r=0.3845:+2 i.63 90 r.0.325 eo E ~ 0.0.0006 70 ~ Z 60 O 50 Z 40 a Q 30 Ja 20 0 0 0 5 10 15 20 25 30 35 40 45 50 55 60 PACK-YEARS SMCKING HlSTCRY rC r=2.4 i 1..322.6 900 .0.2 5 7 800h E I a 700 F 600 ~ 500 0 U 400 < 0.0.008 y300F G .. . a 200 100 04 0 5 10 15 20 25 30 35 40 45 50 55 60 PACK-YEARS SMOK7JG HISTORY . FIG. 4, Relationship between pack-years smoking history and blood levels of carboxyhemoglobin. nicotine or counine. The graphical representation of these relationships includes: the equation for the inserted least-squares best fit line: the correlation coefficient (r); and the level of significance for the correlation. The data are given for smokers smoking filter cigaretsesN'D with nicotine vields of 0.28 to 1.10 mg/cigarette (n= 108). ~ ~ first 4 groups of smokers (Fig. 4). Increased cumulative smokingW history was significantly associated with increased blood concen-jon trations of carboxyhemoglobin• nicotine and cotinine. These,,:, significant correlates were probably due in part to the significant~ 00

I I I I 1 I I I 1 I I I I I I I I I BRIDGES ET AL.. TABLE 9 PREDICTORS OF SERUM THIOCYANATE AND BLOOD CARBOXYHEMOGLOBIN I~ SMOKERS BY MULTIPLE REGRESSION ANALYSES' Thiocvanate- Standard Carboxyhemogiobin; Standard Coefficient Error P Coefficient Error P Intercept 2.68767 1.-17'_85 Independent Variables: A. Cigarettes Past 24 Hours - .5119 0.1044 0.0001 B. Years Smoked - - C. Nicotine Yield - .1248 0.1237 0.32 D. Coffee Consumption .0640 0.0624 0.31 .0067 0.0092 0.47 E. Alcohol-Present .0255 0.0180 0.16 F. -Cumulative - .000026 0.000014 0.07 G. Body Weight -.0118 0.004-4 0.009 -.0158 0.0036 0.0001 Interaction Terms: A*B - -.0111 0.0041 0.008 A'C . - A*D - 01-" 0.0177 0.42 A*E -.0081 0.0055 0.14 A*F - -.000008 0.000004 0.06 A*G .0030 0.0011 0.01 B*C .0053 0.0039 0.18 B"D -.0012 0.0008 0.1 -1 B*E - B*F - B*G - .00046 0.00016 0.004 *Muluple regresston analysis performed on the homogeneous smoking population with nicotine vteld of 0.28 to 1.10 mrJcigarette (Groups 1-4. n= 108). Serum thiocyanate,blood carboxyhemoglobin, and cigarettes past 24 hours were all entered as their logarithmic transformations. +Total R= for thiocvanate was .1164 (p=0.1'_) for the eight vatiable model including interaction tetTtts, tTotal R= for carboxyhemo¢lobin was .2996 (ps0.0001) for the e:ght variable model including interactton terms. ' relationship between pack-years smoking history and normal cigarette consumption per day (r s.704, p= 0.0001) or the num- ber of cigarettes smoked in the last 24 hours (r=.503. p= 0.0001). Unlike pack-years. the number of years smoked did not correlate significantly with cigarette consumption per day (r= .134, p=0.17) or in the past 24 hours (r=.098, p=0.31). However, the number of vears smoked still correlated si¢nifi- cantly with plasma nicotine (r=.268, p=0.005). marginally with carboxyhemoglobin (r=.176. p=0.07), but not with cotinine (r=,106, p=0.:8). Multiple Linear Regression AnalYsis Multiple linear regression analyses were done to determine the most important variables contributing to the prediction of the blood concentrations of smoke constituents (Tables 9 and 10). A better prediction was obtained if the blood concentrations were expressed as a logarithmic vs. linear transformation and if the population utilized was the more homogeneous population (Groups 1-t) according to nicotine vield vs. the total population. Although the independent variables adding significantly to the prediction were ¢enerallv not different in the total vs. homogeneous popula- tions.~ alcohol consumption (or its interaction with the other independent variables) was always more significant in the total vs. homogeneous populations. Interaction terms also improved the prediction over models not containing these interaction terms. Finall.. the inclusion of body mass index did not appreciably affect the model as compared to those models excluding this independent variable. The model not including interaction terms demonstrated that cigarettes smoked in the past 24 hours was the only significant (p=0.083) predictor accounting for 6.28°c of the variation in serum thiocyanate concentration (data not shown). In the model including interaction terms (Table 9). body weight and the product of cigarettes past 24 hours and body weight were significant predictors accounting for 11.64% of the variation in serum thiocyanate: however. the overall re_?ression was still not signifi- cant (p=0.12). ~ Noninteractive regression analysis with carboxyhemoglobin concentration as the dependent variable revealed that cigarettes past 24 hours and body weight were significant predictors account- ing for 17.70% of the variation (data not shown). Although cigarettes past 24 hours and body weight still remained as significant predictors in the interactive model (Table 9). the products of cigarettes past 24 hours and years smoked, and years smoked and weight also contributed significantly to the prediction. Also it should be noted that cumulative alcohol and the product of cigarettes past 241 hours and cumulative alcohol were marginally significant predictors of carboxyhemoglobin levels. Thus, the regression analysis including the independent variables and their interactions accounted totally for =9.9VIc of the variation in carboxvhemos1obin levels. The three- re2ressors which were found to be significant predictors of plasma nicotine concentrations in the noninteractive ~

I i I I I I I I 1 I I I I 1 I I I DETFRMINANTS OF S.\lOI:.E EXPOSLZ2E T..%BLE 10 PREDICTORS OF PL.~SAi.~ NICOTIN`E COTI':1\.E IN SMOKERS BY ML'LTIPLE REGRESSION A.':ALYSES' Coefncier.: Nicoune- Standard Error P Coefficient Cotinine- Standard Error p Interceot 4.0670': 6.40525 Independent Variabies: A. Cigarenes Past 2s Hours B. Years Smoked - C. Nicotine Yield - 1.5555 0.4747 0.001 1.0179 0.5474 0.07 D. Coffee Consumption - -.3134 0.1262 0.01 E. Alcohol-Present -.101' 0.0396 0.01 -.1150 0.0489 0.02 F. -Cumulative .000054 0.000028 0.06 .00011 0.00004 0.006 G. Body Weight - .0344 0.0106 0.002 Interactton Terms: A*B A•C .=941 0.1438 0.0001 A*D - .0948 0.0393 0.02 A•E .0226 0.0108 0.04 .0327 0.0146 0.03 A"F -.00001 0.000008 0.10 -.00004 0.00001 0.002 A'G - .00-tG 0.0011 0.0004 .0055 0.0023 0.02 B'C -.0373 0.0277 0.18 B*D - B*E .0005 0.0003 0.009 B*F .0000005 0.0000005 0.26 B"G .00046 0.00034 0.18 •Multiple regression analysis performed on the homogeneous smoking populatton with nicotine yield of 0.28 to 1.10 mgicigarene (Groups 1-4. n= 10?i. Plasma nicotine and cotinine concentrations and cigarettes past 24 hours were all entered as their loganthrru: tnnsfottnattons. =Total R= for plasrna nicotine was .341" (p:50.0001 i for the eight variable model including interaction terms. :Total R= for plasma cottnine was .3981 (ps0.0001) for the twelve variable model includmg interaction terms. model included cigarettes past 24 hours. years smoked and body weight accounting for 26.05c'rc of the variation (data not shown). Multiple regression analysis using interaction terms produced a markedly different model increasing the total prediction to 31.17Crc and included significant contributions to the prediction by nicotine yield. present alcohol consumption, cigarettes past 24 hours. body weight, and vears smoked either as independent variables or within interaction terms (or both). The positive and highly significant coefficient of cigarettes past 24 hours times nicotine yield suggests that the product of these parameters is an important predictor of plasma tticotine. The product of these parameters is greater in magmtude than nicotine yield itself which may account for the negative but significant coefficient for nicotine yield as a significant independent variable. Present alcohol consumption and body weight as products with cigarettes past 24 hours and present alcohol consumption as an independent variable were also signif- icant predictors. The negative coefficient for the product of cigarettes past 24 hours and bodv weight is likely due to the negative relationship between bod~• weight and plasma nicotine concentration. The product of years smoked and present alcohol consumption was also a significant predictor. Thus. plasma nicotine concentrations are not onlv dependent upon cigarettes past 24 hours, bod% • weight and years smoked, but also nicotine yield and present alcohol consumption. ' Noninteractive. multiple regression analysis with plasma coti- nine as the dependent variable demonstrated that cisarettes past 24 hours was a highly significant (p<0.0001) predictor with margin- ally significant contributions to the prediction by nicotine yield (pz0.12) and body weight (p=0.06) with these variables ac- counting for 29.28% of the variation in plasma cotinine (data not shown). The prediction of plasma cotinine concentrations was markedly improved by regression analysis including the interac- tive terms (39.81cic) (Table 10). In this interactive model, ciga- rettes past 24 hours itself was no longer a significant predictor. however, the products of cigarettes past 24 hours and coffee consumption, present or cumulative alcohol consumption, and body weight were sienificant predictors and replaced cigarettes past 24 hours as an independent variable in the model. With the inclusion of these interactive terms, coffee consumption, present and cumulative alcohol consumption were individually added to the model. lising the interactive model, body weight became a significant (p=0.002) predictor while nicotine yield remained as a marginally significant predictor (p=0.07). DISCUSSION Blood concentrations of thiocyanate. carboxyhemoglobin. nic- otine and cotinine have been used as measures of cigarette smoke intake and absorption. which are dependent upon the number of cigarettes smoked per day. the yield of the ciearette, individual smoking behavior. inhalation behavior, and the uptake of smoke constituents (16). Other factors contributing to the variation in the blood concentrations of these smoke constituents include their rates of absorption. metabolism and excretion. body weight. and their availability from other environmental sources. The purpose of this study was to determine the relationship, if any, between

1 I ~ I I I I I I I I I I I I I I I I 26 blood concentrations of these smoke constituents and population characteristics or smoking history. The subjects in this study were relatively voung and the asvmptomatic smokers had a relatively brief smoking history. Variations in population characteristics were intimately associated with the smoking history variables (e.g., age was significantly related to vears smoked, pack-years smoking history, and the nicotine yield). Differences were observed in subgroups of the smoking population according to the nicotine yield of their cigarette necessitatino the examination of relationships between parameters by regression analyses in the more homogeneous population smoking filtered cigarettes (Groups 1-t. nicotine yield 0.28-1.10 mg). Thus. the biases introduced by the observed differences in population characteristics or smoking history in the total smoking population were minimized. The smokers had significantly higher blood concentrations of thiocyanate. carboxyhemoglobin. nicotine and cotinine than age- matched nonsmokers. Consistent with previous reports (14. 18. 31. 52. 53. 55), considerable overlap of values for thiocyanate and carboxvhemogiobin levels were observed in the smoking and nonsmoking populations. In contrast, plasma nicotine and coti- nine, being more specific for tobacco consumption, allowed for the differentiation between smokers and nonsmokers. Linear correlations amone serum thiocyanate, blood carboxyhemoglobin. and plasma nicotine and cotinine in smokers were consistent with their relative half-lives and specificity for tobacco smoke expo- sure. Further, although passive smoking has been suggested to elevate blood levels of nicotine (40.42), cotinine (20.35). and carboxyhemogiobin (41), we could not demonstrate elevations of blood concentrations of any of these smoke constituents associated with passive smoking or recent marijuana smoking in either smokers or nonsmokers. The log of the cigarette consumption in the past 24 hours was the best predictor of plasma cotinine, followed in order by blood carboxyhemoglobin. plasma nicotine and thiocyanate. The better correlation of cigarette consumption in the past 24 hours with cotinine concentrations is likely due to its specificity for tobacco smoke exposure, its half-life of 19 hours (9). and the fact that its concentration varies less throughout the smoking day than does nicotine and carboxyhemoglobin (8.9). The lesser prediction of nicotine and carboxyhemoglobin concentrations were likely due to their shorter tetminal half-lives (100-120 minutes and 2-4 hours, respectively) (6,44), their variation during the smoking day (8), and the lack of specificity of carboxyhemoglobin for smoking (12.23). Further. individual differences in eliaunation of nicotine and cotini,ne due to urinary pH (3. 5. 9) and carboxyhemoglobin by physical activity (54) have been observed and might have confounded their relationship to 24-hour cigarette consumption. The relatively long half-life of thiocyanate (i.e.. 10-14 days) (18.36) and its poor specificity for tobacco consumption (17, 18. 56) are probably the most important factors contributing to its poor correlation with 24-hour cigarette consumption. In agreement with a previous study (9). while 24-hour cigarette consumption is the best predictor of plasma cotinine, it still accounted for only 13.9% of its variation. suggesting that the number of cigarettes smoked per day is inadequate to explain daily nicotine intake. Although individual differences in rates of inetab- oiism or excretion may partially account for the variability in plasma cotinine. evidence is presented here that yet other factors such as nicotine yield of the cigarette. cumulative smoking history. body weight and alcohol or coffee consumption might also contribute to this variation. In the present study, several observations support the conclu- sions that nicotine yield is an important determinant of smoke exposure and that smokers of low yield cigarettes partially compensate for these low yields by a mechanism other than BRIDGES E7 .-1L. increasing daily cioarette consumption. First. although the mean concentrations of nicotine and cotirune were not signiticantly different in groups of smokers according to the nicotine yield of their cioarette. a decrease in piasma nicotine and cotinine concen- trations was obsen,ed with dec.~reasing nicotine yield. However, the decrease in plasma nicotine or cotinir.e concentrations was not comparable to the decrease in nicotine yield of the cigarette. Secondlv. a siRnificant linear correlation was observed ber•veen nicotine yield and plasma cotinine. «0hile only a marginally significant linear correlation was observed between nicotine yield and plasma nicotine. Finally, multiple regression analyses re- vealed that nicotine yield was a significant predictor of plasma nicotine and a marginally significant predictor of plasma cotimne. Although some investigarors (19. __. 43. 46) have reported that compensation is achieved by increased daily cigarette con- sumption. we, like others (47). did not find this to be the case. The smokers in groups according to nicotine yield smoked approxi- mately the same number of cigarettes daily with no signiftcant linear correlates between nicotine yield and daily ci¢arette con- sumption. Thus, compensation by smokers of low yield cigarettes was likely a result of differences in puffing or inhalation behavior, or both. Although lower mean blood levels of thiocvanate and cart+oxy- hemoQlobin were most notable in the ~:oup of smokers smoking the lowest yield ci¢arettes. these differences were not statistically significant. Further. nicotine vield did not siQttificantly contribute to the prediction of either thiocyanate or carboxyhemoglobtn in multiple regression analyses. These data suggest that smokers of low nicotine yield cigarettes. while possibly reducing their intake of particulate matter ti.e., tar and nicotinet, do not significantly reduce the intake of the gas phase components oi cigarette smoke. Sieniftcant correlations were observed between indices of cumulative smoking history (i.e.. pack-years and years smoked) and blood concentrations of smoke consutuents. Correlations with pack-years was likely due in pan to the relationship between pack-years and daily cigarette consumption. However. years smoked did not correlate siQnificantlv with dailv ciQarette con- sumption but correlated significantly with plasma nicotine and margtnally with carboxyhemogiobin concentrations. Further. vears smoked as an interactive term was a signiticant predictor of carboxyhemoglobin and plasma nicotine in multiple regression analyses. The relationship between years smoked and blood levels of nicotine and carboxyhemoslobin may have been due to a greater cigarette consumption immediately prior to vettipuncturY (not measured in this study) by smokers with a longer cumulative smoking history or alternatively. the development of a tticotine tolerance with cumulative smoking history (13,26). Correlations between blood concentrations of smoke constitu- ents and population characteristics revealed that decreased levels of carboxyhemoglobin, nicotine and cotmiae were associated with increased body weight and (less signuicantly) body mass index. This observation was supported by the fact that body weight (either itself or as an interaction with daily cigarette consumpdon or years smoked) was a significant predictor of carboxyhemoglo- bin, nicotine and cotinine by multiple regression analyses. De- creased concentration of these blood levels of smoke consntuents with increased weight is likely due to the association of volume of distribution with body weight. Significant correlations were also observed between daily cigarette consumption and consumption of alcoholic and caffeine- containing beverages in this, as well as previous studies 1.34). Although an enhanced rate of cigarette smokin2 was associated with ethanol consumption by alcoholics t25), the effects of coffee consumption on the rate of cigarette consumption is equivocal (32-34). In the present study. coffee and alcohol consumption were also signiftcantly correlated wit'rt plasma nicotine and coti-