- Email: [email protected]
Thiocyanate as an indicator of tobacco smoking
PREVENTIVE
MEDICINE
8,
351-357 (1979)
Thiocyanate
as an Indicator
DIETERBORGERS Institute
of Tobacco
Smoking]
AND BURCKHARDJUNGE
of Social Medicine and Epidemiology, Berlin, West Germany
Federal Health
Office,
This study demonstrates the suitability of serum thiocyanate as an indicator of tobacco smoke. The median serum level of smokers rises steadily from 51 pmol/liter in nonsmokers to 174 pmokliter in smokers of more than 20 cigarettes/day. The method can be used in epidemiologic studies of smoking behavior and smoking cessation. To increase the accuracy of predicting smoking habits, this test can be used in conjunction with a determination of expired carbon monoxide.
INTRODUCTION
Smoking is the main source of carbon monoxide (CO) uptake in the general population. This is shown either by determination of carboxyhemoglobin in the blood or, more conveniently for epidemiologic studies, by analysis of CO in expired air (4). The expired air method is rapid, inexpensive, and does not have the technical problems involved in analyzing blood for carboxyhemoglobin. The CO level in the expired air is related directly to the blood carboxyhemoglobin concentration (2). A disadvantage of this method is the short biologic half-life of about 4 hr. Recency of exposure is, therefore, a critical variable in epidemiologic studies using this method. The concordance of CO levels to smoking is so close that in the general population one can use it as a test of habitual smoking. It can be used to assess smoking objectively as compared with the questionnaire as a subjective method. Aside from the measurement of CO, there is a second biochemical test available which is the determination of thiocyanate (SCN). The serum level for SCN derived from cyanide components in tobacco smoke is also a specific, simple, and cheap method to assess smoking habits. The following data and analysis show the suitability of this method in an epidemiologic context. METHODS
A smoking questionnaire with eight questions concerning smoking history was administered to about 1200 persons, who came in mainly for pre-employment examinations (3). The population included both sexes and all ages from 15 to 70 years. Blood was taken and analyzed for the serum SCN level. The determination was done with a photometric method, described by Butts et al. (l), together with other clinical parameters on a multichannel autoanalyzer. The method is cheap, efficient, and fulfills clinical standards of precision, validity, and linearity for an automatic method (3). * hesented at a Workshop American Health Foundation October 10-12, 1978.
on Carbon Monoxide and Cardiovascular Disease, sponsored and the Federal Health O&e, Federal Republic of Germany,
by the Berlin,
351 0091-7435/79/03035 l-07$02.00/O Copyright @ 1979 by Academic hess, Inc. All rights of reproduction in any form reserved.
352
BORGERS AND JUNGE
RESULTS
Figure 1 shows the main results with cigarette smokers as the most important group. Actual smoking was divided into five classes by the questionnaire: nonsmokers, l-4, 5- 10, 11-20, and more than 20 cigarettes per day. The corresponding median serum SCN values are 51 FmoVliter for nonsmokers, rising steadily to 174 I*.mol/liter in smokers of more than 20 cigarettes per day. In addition to the median in each smoking category, the figure shows the central 95% range. One would expect at first that there should be as little overlap as possible between the SCN distributions of the smoking categories, so that persons could be classified by their SCN values into a corresponding questionnaire category without error. One must, however, bear in mind that the questionnaire is not an absolute standard and that the overlapping distributions are partially due to “wrong” questionnaire results with regard to true exposure, as well as from errors in the SCN method itself. Figure 2 shows the results with regard to other smoking habits than cigarettes. Cigarette smokers in total have a median of 141 PmoYliter, cigar smokers have a median of 165 PmoVliter, pipe smokers have a median of 81 PmoVliter, and smokers of two or three types have a median of 130 PmoVliter. The other types (cigar, pipe) are not broken down by amount of smoking because of insufficient numbers.
zoo
100
0 l-4
S-10
11-20
>
20
CismtdDay
FIG. 1. Median and central 95% range of serum thiocyanate values in nonsmokers and cigarette smokers with regard to questionnaire category.
WORKSHOP: CARBON MONOXIDE
AND CVD
353
Swum-Thiocyanate
2ol
1lU
n-395
c
Cigaretta Nonsmokers
Cigar.
Pipe
Combination-
smokers
FIG. 2. Median and central 95% range of serum thiocyanate values in nonsmokers and smokers with different smoking habits.
The different levels, therefore, are due either to different average amounts of smoked tobacco or to different absorption of cyanide with regard to smoking habits. Both questionnaire and SCN determination try to assess actual smoking habits. Actual smoking habits may be reasonably defined as the average amount of cigarettes or other tobacco products smoked during the last week or last month. From experience with a few people who gave up smoking totally at once, we could determine a biologic half-life of about 2 weeks. The SCN level is, therefore, relatively independent from short-term daily variations of smoking behavior. Our questionnaire asked whether a smoker had smoked within the last 3 hr before blood sampling. Figure 3 shows that, within each class, those who had smoked within the last 3 hr had higher SCN medians than those who did not. However, the difference is nearly absent in the highest class with more than 20 cigarettes. The interpretation of these results is difficult, but can be attributed to questionnaire problems as follows: Within each questionnaire category those who smoked in the last 3 hr before blood sampling are those who smoke more, as recency of smoking is correlated to amount of smoking. This can be seen from the steadily rising percentage of smokers within each category during the 3 hr before blood sampling from 10% in the lowest to 65% in the highest class. Furthermore,
354
BORGERS AND JUNGE
Smoking Cigarettes/Day
before
during
72 (38 - 158) ” = 43
5-
10
11 -20
20
Yes
98
104
121 (63 - 196) ” = 25
141
26
(71 - 107) n=5
(41 - 234) n = 51
(51 - 274) ” = 70 1
Difference of the median
“0
l-4
3 hours
blood-sampling
157
17
16’
I87 - 226) n = 77
171
173
(99 - 240) n=R?
(109 - 2411 n = 69
2
* p < 0.05
FIG. 3. Median and central 95% range (in brackets) of serum thiocyanate values in relation to smoking before blood sampling unit of measure: pmoliliter.
the questionnaire asked whether smoke was inhaled or not. We could not find a significant difference in the SCN level with regard to inhalation. Accuracy and Validity of the Test The accuracy of the SCN method as a measure of exposure to tobacco smoke can be tested against questionnaire data only in a relative way. It is possible that SCN gives a better prediction of actual smoking behavior than does the questionnaire, as there is a systematic bias towards underreporting in questionnaire responses. If we take questionnaire data as a yardstick, however, and try to classify persons by their SCN values with regard to smoking, a SCN level of 100 PmoYliter would discriminate in an optimal way. Figure 4 shows the cumulative frequencies of SCN values in each questionnaire category. If we take the smokers of more than 20 cigarettes, we will find 3% erroneously classified as nonsmokers by a level of 1OO~moYliter as a criterion. The error would be larger, of course, in the lower smoking categories. From a practical point of view, there are two different situations where one would use the test: (a) an individual therapeutic situation where one could see the “normalization” of the SCN level as a success of smoking cessation; and (b) a mass trial, i.e., an intervention trial in atherosclerosis and coronary heart disease. One would be able to assess objectively the diminishing smoking habits in a population by a reduction of the average level of SCN in the same way as for cholesterol or blood pressure. The “error” of the SCN test would be critical in the first example; whereas in an epidemiologic situation there are no other sources of SCN increase, which are as massive as smoking and could influence averages in the general population. These other external sources can be seen in Fig. 5. It shows metabolism and
WORKSHOP:
CARBON
MONOXIDE
AND
CVD
355
Cumulnb Fmq”ancie3 ISI
FIG. 4. Cumulative frequency curves for nonsmokers and different smoking categories.
sources of SCN. External sources other than smoking are cyanide-containing products, industrial cyanide in the metal industry, cyanide-producing bacteria, or SCN itself in cabbage, mustard, turnips etc. Cyanide in tobacco smoke and from other sources is converted mainly by the liver enzyme sulfur transferase to the nontoxic SCN, which is excreted by urine. These other sources are possible reasons for error with regard to smoking. To increase the accuracy one may use this method in conjunction with a determination of expired CO.
FIG. 5. Uptake, intermediate metabolism, and excretion of thiocyanate.
356
BORGERS AND JUNGE Mean
Mean
Serum-Thiocyanata
Expired
CO
11
200
20
150
15
100
10
50
5 -
0
l-9 Number
FIG.
-
10 - 20 of Cigamttm
-
Expired
CO
>20
per Day
6. Mean expired CO and mean serum thiocyanate levels based on a figure in reference (5).
Our study does not have data on CO levels. Therefore, to compare both methods, we used data from a study of Vogt, et al. (5). Figure 6 shows mean expired CO and mean SCN levels with regard to questionnaire responses: nonsmokers, 1-9, 10-20, and more than 20 cigarettes. The curves have the same direction, with a steeper slope for CO. Figure, 7 shows the correlation coefficients of expired CO, SCN, reported number of cigarettes, and time since last cigarette before the test. The correlation between CO and SCN (Y = 0.571) is greater than the correlation between either test and questionnaire responses. The coefficient between time since last cigarette and CO is larger (r = -0.416) than between time since last cigarette and SCN (Y = -0.255). This is the consequence of the different biologic half-life of CO and SCN. In their study Vogt et al. (5) demonstrated an increase of accuracy in predicting smoking behavior using both methods simultaneously.
j;.l FIG.
7. Pearson correlation coefficients among smokers based on a figure in reference (5).
WORKSHOP:
CARBON
MONOXIDE
AND
CVD
357
DISCUSSION The determination of serum SCN and the determination of CO in expired air are both valid methods to assess personal smoking habits. The serum SCN method can be used either alone or, if more accuracy is needed, in conjunction with expired CO. The sources of error (analytical, other sources of SCN) are small enough to permit the use in an epidemiologic study of smoking exposure and smoking reduction. REFERENCES 1. Butts, W. C., Kuehneman, M., and Widdowson, G. M. Automated method for determining serum thiocyanate, to distinguish smokers from nonsmokers. C/in. Chem. 20, 1344 (1974). 2. Jones, R. H., Ellicott, M. R., Cadigan, J. B., and Gaensler, E. A. The relationship between alveolar and blood carbon monoxide concentrations during breath-holding: Simple estimation of COHb-saturations. J. Lab. Clin. Med. 51, 553 (1958). 3. Junge, B., Borgers, D., Berkholz, M.-B., Thefeld, W., and Hoffmeister, H. Thiocyanat im Serum als Indikator fur die SchadstofIbelastung durch Tabakrauch. Arbeirsmed. Sozialmed. Praventivmed. 13, 13 (1978). 4. Ringold, A., Goldsmith, J. R., Helwig, H. L., Finn, R., and Schuette, F. Estimating recent carbon monoxide exposures: A rapid method. Arch. Environ. Health 5, 308 (1962). 5. Vogt, T. M., Selvin, S., Widdowson, G., and Hulley, S. B. Expired air carbon monoxide and serum thiocyanate as objective measures of cigarette exposure. Amer. J. Public Health 67,545 (1977).
MEDICINE
8,
351-357 (1979)
Thiocyanate
as an Indicator
DIETERBORGERS Institute
of Tobacco
Smoking]
AND BURCKHARDJUNGE
of Social Medicine and Epidemiology, Berlin, West Germany
Federal Health
Office,
This study demonstrates the suitability of serum thiocyanate as an indicator of tobacco smoke. The median serum level of smokers rises steadily from 51 pmol/liter in nonsmokers to 174 pmokliter in smokers of more than 20 cigarettes/day. The method can be used in epidemiologic studies of smoking behavior and smoking cessation. To increase the accuracy of predicting smoking habits, this test can be used in conjunction with a determination of expired carbon monoxide.
INTRODUCTION
Smoking is the main source of carbon monoxide (CO) uptake in the general population. This is shown either by determination of carboxyhemoglobin in the blood or, more conveniently for epidemiologic studies, by analysis of CO in expired air (4). The expired air method is rapid, inexpensive, and does not have the technical problems involved in analyzing blood for carboxyhemoglobin. The CO level in the expired air is related directly to the blood carboxyhemoglobin concentration (2). A disadvantage of this method is the short biologic half-life of about 4 hr. Recency of exposure is, therefore, a critical variable in epidemiologic studies using this method. The concordance of CO levels to smoking is so close that in the general population one can use it as a test of habitual smoking. It can be used to assess smoking objectively as compared with the questionnaire as a subjective method. Aside from the measurement of CO, there is a second biochemical test available which is the determination of thiocyanate (SCN). The serum level for SCN derived from cyanide components in tobacco smoke is also a specific, simple, and cheap method to assess smoking habits. The following data and analysis show the suitability of this method in an epidemiologic context. METHODS
A smoking questionnaire with eight questions concerning smoking history was administered to about 1200 persons, who came in mainly for pre-employment examinations (3). The population included both sexes and all ages from 15 to 70 years. Blood was taken and analyzed for the serum SCN level. The determination was done with a photometric method, described by Butts et al. (l), together with other clinical parameters on a multichannel autoanalyzer. The method is cheap, efficient, and fulfills clinical standards of precision, validity, and linearity for an automatic method (3). * hesented at a Workshop American Health Foundation October 10-12, 1978.
on Carbon Monoxide and Cardiovascular Disease, sponsored and the Federal Health O&e, Federal Republic of Germany,
by the Berlin,
351 0091-7435/79/03035 l-07$02.00/O Copyright @ 1979 by Academic hess, Inc. All rights of reproduction in any form reserved.
352
BORGERS AND JUNGE
RESULTS
Figure 1 shows the main results with cigarette smokers as the most important group. Actual smoking was divided into five classes by the questionnaire: nonsmokers, l-4, 5- 10, 11-20, and more than 20 cigarettes per day. The corresponding median serum SCN values are 51 FmoVliter for nonsmokers, rising steadily to 174 I*.mol/liter in smokers of more than 20 cigarettes per day. In addition to the median in each smoking category, the figure shows the central 95% range. One would expect at first that there should be as little overlap as possible between the SCN distributions of the smoking categories, so that persons could be classified by their SCN values into a corresponding questionnaire category without error. One must, however, bear in mind that the questionnaire is not an absolute standard and that the overlapping distributions are partially due to “wrong” questionnaire results with regard to true exposure, as well as from errors in the SCN method itself. Figure 2 shows the results with regard to other smoking habits than cigarettes. Cigarette smokers in total have a median of 141 PmoYliter, cigar smokers have a median of 165 PmoVliter, pipe smokers have a median of 81 PmoVliter, and smokers of two or three types have a median of 130 PmoVliter. The other types (cigar, pipe) are not broken down by amount of smoking because of insufficient numbers.
zoo
100
0 l-4
S-10
11-20
>
20
CismtdDay
FIG. 1. Median and central 95% range of serum thiocyanate values in nonsmokers and cigarette smokers with regard to questionnaire category.
WORKSHOP: CARBON MONOXIDE
AND CVD
353
Swum-Thiocyanate
2ol
1lU
n-395
c
Cigaretta Nonsmokers
Cigar.
Pipe
Combination-
smokers
FIG. 2. Median and central 95% range of serum thiocyanate values in nonsmokers and smokers with different smoking habits.
The different levels, therefore, are due either to different average amounts of smoked tobacco or to different absorption of cyanide with regard to smoking habits. Both questionnaire and SCN determination try to assess actual smoking habits. Actual smoking habits may be reasonably defined as the average amount of cigarettes or other tobacco products smoked during the last week or last month. From experience with a few people who gave up smoking totally at once, we could determine a biologic half-life of about 2 weeks. The SCN level is, therefore, relatively independent from short-term daily variations of smoking behavior. Our questionnaire asked whether a smoker had smoked within the last 3 hr before blood sampling. Figure 3 shows that, within each class, those who had smoked within the last 3 hr had higher SCN medians than those who did not. However, the difference is nearly absent in the highest class with more than 20 cigarettes. The interpretation of these results is difficult, but can be attributed to questionnaire problems as follows: Within each questionnaire category those who smoked in the last 3 hr before blood sampling are those who smoke more, as recency of smoking is correlated to amount of smoking. This can be seen from the steadily rising percentage of smokers within each category during the 3 hr before blood sampling from 10% in the lowest to 65% in the highest class. Furthermore,
354
BORGERS AND JUNGE
Smoking Cigarettes/Day
before
during
72 (38 - 158) ” = 43
5-
10
11 -20
20
Yes
98
104
121 (63 - 196) ” = 25
141
26
(71 - 107) n=5
(41 - 234) n = 51
(51 - 274) ” = 70 1
Difference of the median
“0
l-4
3 hours
blood-sampling
157
17
16’
I87 - 226) n = 77
171
173
(99 - 240) n=R?
(109 - 2411 n = 69
2
* p < 0.05
FIG. 3. Median and central 95% range (in brackets) of serum thiocyanate values in relation to smoking before blood sampling unit of measure: pmoliliter.
the questionnaire asked whether smoke was inhaled or not. We could not find a significant difference in the SCN level with regard to inhalation. Accuracy and Validity of the Test The accuracy of the SCN method as a measure of exposure to tobacco smoke can be tested against questionnaire data only in a relative way. It is possible that SCN gives a better prediction of actual smoking behavior than does the questionnaire, as there is a systematic bias towards underreporting in questionnaire responses. If we take questionnaire data as a yardstick, however, and try to classify persons by their SCN values with regard to smoking, a SCN level of 100 PmoYliter would discriminate in an optimal way. Figure 4 shows the cumulative frequencies of SCN values in each questionnaire category. If we take the smokers of more than 20 cigarettes, we will find 3% erroneously classified as nonsmokers by a level of 1OO~moYliter as a criterion. The error would be larger, of course, in the lower smoking categories. From a practical point of view, there are two different situations where one would use the test: (a) an individual therapeutic situation where one could see the “normalization” of the SCN level as a success of smoking cessation; and (b) a mass trial, i.e., an intervention trial in atherosclerosis and coronary heart disease. One would be able to assess objectively the diminishing smoking habits in a population by a reduction of the average level of SCN in the same way as for cholesterol or blood pressure. The “error” of the SCN test would be critical in the first example; whereas in an epidemiologic situation there are no other sources of SCN increase, which are as massive as smoking and could influence averages in the general population. These other external sources can be seen in Fig. 5. It shows metabolism and
WORKSHOP:
CARBON
MONOXIDE
AND
CVD
355
Cumulnb Fmq”ancie3 ISI
FIG. 4. Cumulative frequency curves for nonsmokers and different smoking categories.
sources of SCN. External sources other than smoking are cyanide-containing products, industrial cyanide in the metal industry, cyanide-producing bacteria, or SCN itself in cabbage, mustard, turnips etc. Cyanide in tobacco smoke and from other sources is converted mainly by the liver enzyme sulfur transferase to the nontoxic SCN, which is excreted by urine. These other sources are possible reasons for error with regard to smoking. To increase the accuracy one may use this method in conjunction with a determination of expired CO.
FIG. 5. Uptake, intermediate metabolism, and excretion of thiocyanate.
356
BORGERS AND JUNGE Mean
Mean
Serum-Thiocyanata
Expired
CO
11
200
20
150
15
100
10
50
5 -
0
l-9 Number
FIG.
-
10 - 20 of Cigamttm
-
Expired
CO
>20
per Day
6. Mean expired CO and mean serum thiocyanate levels based on a figure in reference (5).
Our study does not have data on CO levels. Therefore, to compare both methods, we used data from a study of Vogt, et al. (5). Figure 6 shows mean expired CO and mean SCN levels with regard to questionnaire responses: nonsmokers, 1-9, 10-20, and more than 20 cigarettes. The curves have the same direction, with a steeper slope for CO. Figure, 7 shows the correlation coefficients of expired CO, SCN, reported number of cigarettes, and time since last cigarette before the test. The correlation between CO and SCN (Y = 0.571) is greater than the correlation between either test and questionnaire responses. The coefficient between time since last cigarette and CO is larger (r = -0.416) than between time since last cigarette and SCN (Y = -0.255). This is the consequence of the different biologic half-life of CO and SCN. In their study Vogt et al. (5) demonstrated an increase of accuracy in predicting smoking behavior using both methods simultaneously.
j;.l FIG.
7. Pearson correlation coefficients among smokers based on a figure in reference (5).
WORKSHOP:
CARBON
MONOXIDE
AND
CVD
357
DISCUSSION The determination of serum SCN and the determination of CO in expired air are both valid methods to assess personal smoking habits. The serum SCN method can be used either alone or, if more accuracy is needed, in conjunction with expired CO. The sources of error (analytical, other sources of SCN) are small enough to permit the use in an epidemiologic study of smoking exposure and smoking reduction. REFERENCES 1. Butts, W. C., Kuehneman, M., and Widdowson, G. M. Automated method for determining serum thiocyanate, to distinguish smokers from nonsmokers. C/in. Chem. 20, 1344 (1974). 2. Jones, R. H., Ellicott, M. R., Cadigan, J. B., and Gaensler, E. A. The relationship between alveolar and blood carbon monoxide concentrations during breath-holding: Simple estimation of COHb-saturations. J. Lab. Clin. Med. 51, 553 (1958). 3. Junge, B., Borgers, D., Berkholz, M.-B., Thefeld, W., and Hoffmeister, H. Thiocyanat im Serum als Indikator fur die SchadstofIbelastung durch Tabakrauch. Arbeirsmed. Sozialmed. Praventivmed. 13, 13 (1978). 4. Ringold, A., Goldsmith, J. R., Helwig, H. L., Finn, R., and Schuette, F. Estimating recent carbon monoxide exposures: A rapid method. Arch. Environ. Health 5, 308 (1962). 5. Vogt, T. M., Selvin, S., Widdowson, G., and Hulley, S. B. Expired air carbon monoxide and serum thiocyanate as objective measures of cigarette exposure. Amer. J. Public Health 67,545 (1977).