Tobacco Documents Online

9 I. CHEMICAL CHARACTERIZATION OF TOBACCO SMOKE FROM BEAGLE DC1G INHALATION EXPOSURE SYSTEMS R. B. Quincy and A. H. Marshall Introduction. While monitoring of inhalation bioassay exposures seeks to define chemically the nature of the smoke offered the animals at the expo- sure laboratory, there is an important need to define the exposure systems off-site, at ORNL. Because of differences in smoking protocols, etc., at the various exposure laboratories, characterization studies performed at ORNL can act as a standard reference point for interlaboratory comparison. Under tightly controlled environmental conditions at ORNL, cigarettes can be weight and resistance-to-draw selected for these studies. Air flow is carefully regulated. Also, sampling and analysis methods which are either too complicated, time consuming, or still in the development stage can be more easily applied at ORNL. Unanticipated problems can be dealt with before proceeding. For these reasons, we have sought to characterize the output of the various dog exposure systems which have been used by NCI as to their production and delivery of total particulate matter, tar, nicotine, carbon monoxide, carbon dioxide, and hydrogen cyanide. Results of sampling and validation studies, and comparison of the various systems are reported be- low. Method Validation. It is a much more difficult task to obtain a meaningful smoke sample from the exit of the cannula of a dog exposure system than from a conventional analytical smoking machine. Because of the need to mimic the exposure situation, smoke must be withdrawn from the cannula and stand tube by a relatively high velocity, pulsed air flow. This pulsed withdrawal more closely resembles the breathing pattern of the dog than would

10 a slower, more continuous withdrawal. Effective sampling for chemical analysis must address the problem of trapping the specific constituent under high air flow. Presently, to mimic the animal breathing, we employ a Harvard Large Animal Ventilator (respirator) set at 16 X 100 ml strokes per minute. Last year we reported on sampling methods for exposure system delivery of carbon monoxide and carbon dioxide. This year we,discuss sampling of the cannula exit for hydrogen cyanide. Whole smoke hydrogen cyanide is usually distributed between the particu- late and vapor phases. Under the flow conditions established by conventional analytical smoking, usually about half of the HCN is found in the particulate matter, while the other 50o is found in the vapor phase. For chemical analy- sis, the HCN in the vapor phase is collected immediately downstream of the standard Cambridge Filter Pad in a silica gel trap. Both the filter pad and silica gel are extracted with sodium hydroxide solution and the HCN is converted to cyanogen chloride. A colored complex with pyridine and a pyra- zolone derivative is formed, and the absorbance is measured spectrophotometrically. The chief difficulty associated with determining the HCN delivered to the cannula exit is the volatilization of the HCN out of the particulate matter due to the high, pulsed air flow. It is essential that this HCN now in the gas phase be trapped on the silica gel. If the silica gel trap were not being effective at retaining the HCN, presumably the total amount of HCN trapped at the cannula exit would be reduced as the air flow through the system increased. Table 1-2 compares the amounts of HCN trapped at the cannula (on both the filter pad and silica gel trap) as a function of the time which the Respirator remains running following the last puff of the cigarette. Statistically, the 180-second HCN value is slightly lower than ~; the 60-second value (-. 3%), but is not different from the 30-second value. ~ On several of these experiments, a second silica ~ gel trap was installed CA a CD

11 TABLE 1-2 Cannula Exi~ Hydrogen Cyanide Deliveriesa Influence of Air Flow Through Trapping System on Amount Collected (Mean ± one standard deviation) Time of Respir Last Puff of C ator Run Past Amount of HCN Collected igarette (sec} (pg/cigarette) 30 318 -F 11 60 325 ± 6 180 314 ± 4 aCode 32 cigarettes smoked on ADL/II exposure system at 2 puffs per minute. bHarvard respirator set at 16 X 100 cc strokes per minute.

12 immediately downstream of the first one. No detectable levels of HCN could be found in this second trap. From these data, it was concluded that all of the HCN reaching the exit of the cannula was in fact being trapped. Analyses of individual filter pads and traps indicated that, in contrast to an analtyical sample, only 15% of the total HCN was found on the Cambridge filter, while 85% was retained in the silica gel trap. Comparison of Exposure Systems. The purpose of this work was to charac- terize the dog inhalation exposure systems as to their smoke generation and delivery capabilities for one particular NCI variant. The variant chosen for the study was a Series IV Code 32, a Standard Experimental Blend cigarette. The three exposure systems have been used in four bioassay lab- oratories over the past several years. It is important to place that use in perspective. Both the ADL/I and ADL/II systems were developed by Arthur D. Little, Inc. The ADL/I was the earlier model, and was used in the VA Hospital dog bioassay. The system was used with an uncuffed (hard) cannula and, because it was not highly efficient in delivering smoke particulates to experimental animals, the system smoked cigarettes at 3 puffs per minute, in order to offer a relatively high smoke dose to the animal. The ADL/II system is a newer generation model, and, while conceptually very similar to the ADL/I, important improvements in design caused it to be much more efficient in smoke particulate delivery. It has been used with cuffed (inflatable balloon) cannulas only, smoking cigarettes at 2 puffs per minute. The BNW system was developed by Battelle's Pacific Northwest Laboratories for to- bacco smoke exposure of dogs.' Conceptually, it offers certain advantages over the ADL systems, in that there is very little "manipulation" of the smoke before it's inhaled. However, to achieve conceptual simplicity, the :r

13 machine is mechanically complex. It was used with an uncuffed cannula, and smoke particulates which were not inhaled by the animal were left to coat the inner workings of the machine. The BNW system smoked cigarettes at two puffs per minute, and was used to chronically expose dogs for a total of more tha'n 300 days at two different laboratories. It is no longer in use in any NCI chronic exposures. Since the amount and "quality'° of tobacco smoke which emanates from an exposure system is dependednt on the smoke going into the system, the amount and composition of smoke which the exposure system generates is an important consideration in judging comparability of the bioassays. The amount of smoke generated is taken as the machine input. Table 1-3 compares the smoke constituents generated by the three exposure systems with those of smoke gene- rated under analytical conditions. Results of the ADL/I are compared under two puffing regimes, one for comparison with other systems and one to compare with the bioassay conditions under which it was used. Even though the data are a result of many analyses, there are some findings which cannot be easily explained in a self-consistent manner. Thus, the findings should be considered somewhat preliminary. In general, the exposure systems produce significantly less particulate matter and tar than does the analytical system smoking ident- ical cigarettes. All of the exposure systems were checked daily to insure proper puffing performance, but small leaks could occur where the cigarette is press-fitted into the exposure system cigarette holder. It seems unlikely that small leaks could cause such a difference in particulate phase generation.. The ADL/II is quite close to the analytical smoker in the amount of nico- tine generated. Both the BNW`and ADL/I at 2 puffs/min are somewhat higher. The BNW is a so-called free smoker (butt of cigarette open to air during static burning) and usually nicotine levels are somewhat higher under those conditions. The ADL/I at 3 puffs/min produces much less nicotine

TABLE 1-3 Comparison of Exposure System Smoke Generationa with Analytical Smoking Code 32 Cigarette 2 puffs/minute 3 puffs/minute Constituent Analyticalb ADL/I ADL/II BNW Analyticalb ADL/I Total Particulate Matter (mg/cig) 55.2 ± 2.6 40.1 ± 0.4 43.9 ± 0.9 41.4 ± 1.6 65.0 ± 0.8 52.7 ± 2.8 Nicotine (mg/cig) 2.72 ± 0.03 2.99 ± 0.04 2.67 ± 0.11 2.94 ± 0.07 3.85 ± 0.06 3.06 ± 0.21 Tar (mg/cig) . 46.0 ± 2.6 33.3 ± 0.5 36.1 ± 2.1 34.4 ± 0.6 53.1 ± 0.6 41.8 ± 0.2 Carbon Monoxide (mg/cig) 31.1 ± 2.1 NRC NR NR 26.0 ± 0.5 NR Carbon Dioxide (mg/cig) 91.3 ± 1.4 NR NR NR 105 ± 0.5 NR Hydrogen Cyanide (pg/cig) 389 ± 10 359 ± 18 357 ± 31 321 ± 14 430 ± 13 386 ± 9 aUrawn from the cigarette; amount of smoke generated from the cigarette is defined as the machine input, bPhipps and Bird analytical smoking machine cNR: Not Reported. "Input" CO and C0,, values cannot be conveniently obtained for animal exposure systems. zqs1,C4se

15 than does the analytical system at three puffs per minute. Sampling and analysis of whole smoke HCN is often subject to much more variation than are other smoke constituents. While the amount of HCN gene- rated by the ADL/I & II at two puffs per minute is statistically different from analytical smoking, the levels are quite similar. Interestingly, levels of HCN and tar generated by the analytical system at two and three puffs per minute do not seem proportionately greater at the higher puffing rate. Also, the ratio of C():CQ2 is lower at the higher.puffing rate, suggesting that combus- tion processes may he altered, which may implicate other changes in smoke chemistry. Table 1-4 compares the deliveries of smoke constituents of the various exposure systems from the cannula exit. First, it is important to realize that it is very difficult to obtain what could be considered a"typical" cannula exit value of any constituent for the ADL/I. As reported on in previous years, the ADL/I is very susceptible to large changes in smoke particulate phase output depending on the number of cigarettes smoked through it. Every effort was made to sample the ADL/I as reproducibly as possible, but unex- pected results may be due to this dependence on the number of cigarettes smoked since the last cleaning. It appears that the ADL/I delivers more CO and COz than the cigarettes generate under analytical conditions at both 2 and 3 puffs per minute. The delivery of these gas phase constituents are similar at both puffing rates. This finding is somewhat unexpected, but then, the amount of HCN generated by the.two machines at the two puffing rates is also very similar. Nicotine deliveries for all systems at two puffs,/minute are essentially identical. Tar levels for all machines, while not equal, are comparable. The ADL/II delivers much more HCN than the other systems. And interestingly, the ADI./I

TABLE 1-4 Comparison of Exposure System Cannula Exit Smoke Deliveries Code 32 Variant 2 puffs/minute 3 puffs/minute Constituent ADL/I ADL/II BNW ADL/I TPM (mg/ci g) 31.0 ± 0.8 34.9 ± 1.3 36.2 ± 2.2 34.2 ± 1.2 Nicotine (mg/cig) 2.22 ± 0.09. 2.22 ± 0.17 2.28 ~ 0.18 2.52 ± 1.1 Tar (mg/cig) 26.4 ± 0.8 29.0 i- 1.1 30.8 ± 0.3 28.1 ± 0.9 CO (mg/cig) 23.1 ± 2.1 20.4 ± 1.6 19.4 ± 1.5 26.3 ± 0.5 C02 (mg/cig) 114 ± 8 88.6 ± 40 80.7 ~ 4.1 117 ± 2 HCN (-pg/c4g) 191 ± 18 317 ± 10 241 ± 13 204 ± 9 P9S'4C4fif3

17 at 3 puffs/min delivers very little more smoke than at 2 puffs per minute. Table 1-5 compares the apparent delivery efficiencies for all three exposure systems. The ADL/II appears to be the most consistently efficient of the three systems, but the BPiW appears more efficient at delivery of tar. The ADL/I's efficiency appears to drop at the higher puffing rate. That there exist some large system-to-system differences in the levels of certain smoke constituents delivered illustrates the importance of off-bioassay-site characterization of those exposure systems. Clearly, the reasons for these differences are not well understood. We are presently investigating poten- tial causes of these differences. Com arison of Delivered Smokes. The ADL/II-cuffed cannula smoke expo- sure system has become the standard for the NCI beagle dog inhalation bio- assays. Currently, three bioassays have employed this system, and a fourth is scheduled to start soon. As part of our effort to perform careful char- acterization of the smoke emanating from the exposure systems, we have determined some selected constituents in some of the smokes to which animals have been exposed with the ADL/II system. The results to date are reported in Table 1-6. A few observations can be made concerning the results. Two (Codes 04 and 29) of the three SEB IV variants on Table 1-6 have identical tar and nicotine deliveries. CO and CQ2 content of their cannula exit smokes are also quite similar, but HCN deliveries are somewhat different. The other SEB IV variant (Code 14) delivers somewhat more nicotine but a level of tar similar to the other SEB IV's. Interestingly, the two variants which contain some extracted tobaccos (Code"11 and 13) have a much greater HCN concentration

TABLE 1-5 Comparison of Exposure System Smoke Delivery Efficiencies* Code 32 Variant Constituent ADL/I 2 puffs/minute ADL/II BNW 3 puffs per minute ADL/I TPM .77 .80 .87 .65 Nicotine .74 .83 .78 .82 Tar .79 .80 .90 .67 Hydrogen Cyanide .53 .89 .75 .53 *Delivery efficiency defined amount of smoke constituent delivered to cannula exit divided by amount generated at machine input. ~BS~'~~69