Effects of Charcoal on Carbonyl Delivery from Commercial, Research, and Make-Your-Own Cigarettes

Abstract

Previous literature has shown that adding charcoal to cigarette filters can have varying effects on the delivery of toxic carbonyls depending on filter design, amount of charcoal, and puffing profiles. However, these studies have relied on either comparisons between commercially available charcoal and noncharcoal filtered cigarettes or experimental modification of filters to insert a charcoal plug into existing cellulose acetate filters. Make-your-own (MYO) cigarettes can help obviate many of the potential pitfalls of previous studies; thus, we conducted studies using commercial charcoal cigarettes and MYO cigarettes to determine the effects of charcoal on carbonyl delivery. To do this, we analyzed carbonyls in mainstream smoke by HPLC-UV after derivatization with 2,4-dinitrophenylhydrazine (DNPH). Charcoal was added in-line after the cigarettes or through the use of MYO charcoal cigarette tubes.MYO cigarettes had carbonyl deliveries similar to that of 3R4F research cigarette, regardless of tobacco type. The greatest effect on carbonyl delivery was observed with 200 mg of charcoal, significantly reducing all carbonyls under both methods tested. However, “on-tow” design charcoal filters, available on many commercially available charcoal brands, appeared to have a minimal effect on carbonyl delivery under intense smoking methods. Overall, we found that charcoal, when added in sufficient quantity (200 mg) as a plug, can substantially reduce carbonyl delivery for both MYO and conventional cigarettes. As carbonyls are related to negative health outcomes, such reductions may be associated with reductions in carbonyl-related harm in smokers.

Graphical Abstract

INTRODUCTION

Carbonyls are an important class of toxicants in tobacco smoke that have been linked to negative health outcomes such as cancer and respiratory diseases.¹ Seven carbonyls are listed on the Food and Drug Administration (FDA) published list of harmful and potentially harmful constituents (HPHCs): acetaldehyde, acrolein, acetone, formaldehyde, propionaldehyde, crotonaldehyde, and methyl ethyl ketone (MEK).² Previous literature has shown that adding charcoal to cigarette filters can have varying effects on carbonyl delivery depending on filter design, amount of charcoal, and puffing conditions. Preference for charcoal cigarettes varies widely across the globe, and in some countries, such as Japan, charcoal filtered brands have accounted for 85% of total cigarette sales.³ This has led some to speculate that the addition of charcoal could explain at least in part the lower lung cancer rate among Japanese compared to Western smokers due to its ability to lower toxicants such as carbonyls.⁴

In many commercially available charcoal cigarettes which utilize an “in-tow” design and relatively low levels of charcoal (<50 mg), the concentrations of carbonyls in mainstream smoke are similar to those in noncharcoal brands under intense puffing conditions (5–16% reduction) and slightly lower (36–50% reduction) under less intensive puffing conditions.^5,6 However, large reductions in total carbonyls were observed even under intense smoking conditions in commercially available products which incorporate higher amounts of charcoal (100–400 mg) (45–56% reductions)⁵ or when a plug containing these higher levels of charcoal is experimentally incorporated by modification of existing cellulose acetate filters in commercial cigarettes (71– 98% reductions).⁷ Altogether, these studies suggest that larger levels of charcoal (>100 mg) in a plug-style configuration are required for optimal removal of carbonyls.

Although there have been several studies on the effects of charcoal on carbonyl delivery, these have primarily relied on the use of either commercially available charcoal-filtered brands, which contain low levels of charcoal,^6,8,9 or experimentally modified noncharcoal cigarettes with charcoal inserted into the existing cellulose acetate filters.⁷ In the former studies using commercial charcoal filtered cigarettes, comparisons were made with noncharcoal brands which potentially differ by factors other than charcoal (e.g., tobacco blend, wrapping paper design, and ventilation), and direct comparisons with products that are identical in all ways except the presence of charcoal were not possible without modifying the products themselves. Approaches that use experimentally modified filters have the potential of introducing measurement artifacts resulting from the cutting and taping of the filter during the insertion of charcoal and associated blockage of filter ventilation. These cigarettes also are not able to be tested using the International Organization of Standardization (ISO) method as modifying the filter and using tape blocks ventilation, which is not blocked on the ISO method.

Make-your-own (MYO) cigarettes offer the possibility of avoiding many of these potential pitfalls. MYO cigarettes, a subset of roll-your-own (RYO) cigarettes, are made by smokers themselves using premade tubes that are filled with tobacco using a filling machine. The tubes and filters used as well as amount and type of tobacco placed in the cigarette can all be varied by the user; however, this design allows for tight control of these important variables in an experimental laboratory setting. MYO cigarettes are primarily made with cigarette or pipe tobacco with many MYO cigarette smokers choosing pipe tobacco due to its lower cost.¹⁰ Unlike most RYO cigarettes, MYO cigarettes are far more likely to include a filter as most premade tubes on the market typically include a cellulose acetate filter.^11,12 However, these tubes can be of many different shapes and sizes and include varying filter types, including charcoal filters. Thus, in the present studies, we utilized commercially available MYO charcoal and noncharcoal filtered tubes to examine the specific impact of charcoal on carbonyl delivery, while keeping other factors, such as tobacco blend, constant. These studies also allow for the first examination of carbonyl delivery from these products, which have been growing in popularity in the United States.^13,14 We also examined the effects of placing in-line charcoal filters downstream of conventional cigarettes to fully assess the impact of charcoal on carbonyl deliveries without requiring product modification.

METHODS

Materials.

Acetonitrile (ACN) and 12 N hydrochloric acid (HCl) were purchased from Fisher Scientific (Pittsburgh, PA, United States). Diglyme and dinitrophenylhydrazones of formaldehyde, acetaldehyde, crotonaldehyde, propionaldehyde, and MEK were purchased from Sigma-Aldrich (St. Louis, MO, United States). 2,4-Dinitrophenylhydrazine (DNPH) was purchased from BOC Sciences (Shirley, NY, United States), recrystallized in acetonitrile to remove water, and then stored in a vacuum desiccation chamber until use.¹⁵

Cigarettes.

The 3R4F research cigarettes were obtained and shipped from the University of Kentucky (Lexington, Kentucky, United States). These research cigarettes were used as a reference for cigarettes on the United States market, as previous studies have shown that the carbonyl levels delivered by these research cigarettes are comparable to the levels delivered from commercial cigarettes.^16,17 United States branded charcoal cigarettes (Kent HD, Parliament Aqua 5, and Lark) were purchased online (www.ciggiesworld.com). MYO pipe tobacco (The Good Stuff Red) and cigarette tobacco (Rave Red) were purchased locally (Dauphin County, PA, United States). MYO tubes (Cartel Brand, M Tobacco LTD, Plovdiv, Bulgaria) were purchased through eBay to have a MYO king-size tube that came in both a version with a charcoal filter (Carbon Filter 20 mm Filtered Cigarette Tubes) and a version with a regular cellulose acetate filter (Filter Plus 20 mm Filtered Cigarette Tubes). The tubes and MYO tobacco were stored at room temperature (22 ± 1 °C) until use. MYO cigarettes were assembled using a Powermatic 2Plus Cigarette Injector Machine (Zico United States, Inc., Brea, CA, United States; ordered through Amazon). All MYO tubes were weighed before adding the tobacco, and then the assembled cigarettes were weighed. Each MYO cigarette contained 0.82 ± 0.03 g of tobacco. Prior to use, cigarettes were conditioned in a constant humidity chamber (60% relative humidity, 22 ± 1 °C) for 24 to 72 h to ensure similar moisture content across products.¹⁸

Mainstream Smoke Generation.

Mainstream smoke was generated using a Human Puff Profile Cigarette Smoking Machine (CSM-HPP; CH Technologies, NJ, United States). One cigarette was smoked at a time under the International Organization of Standardization (ISO; puff volume: 35 mL, puff duration: 2 s, interpuff interval: 60 s, open ventilation; butt length: tipping paper + 3 mm)^19,20 and Health Canada Intense (HCI; puff volume: 55 mL, puff duration: 2 s, interpuff interval: 30 s, blocked ventilation; butt length: tipping paper + 3 mm)^20,21 methods. We performed three replicates (each consisting of one cigarette) for all conditions unless otherwise mentioned. Per puff yields were determined by dividing carbonyl yields by the number of puffs to smoke an entire cigarette and are not measured for each puff directly. The lighting puff was included in all analyses.

In-Line Charcoal Filter.

To directly test the impact of charcoal on carbonyl delivery, we utilized an in-line charcoal filter placed downstream of either conventional or MYO cigarettes. This filter, which consisted of charcoal contained in a 1 mL Supelco plastic tube (Bellefonte, PA, United States) held in place by deactivated glass wool (Restek, Bellefonte, PA, United States), was placed in-line between the cigarette and the pump (Figure 1). The charcoal used for all studies was Donau Carbon Alcarbon CI 60/30 × 70 (Donau Carbon US LLC, Dunnellon, FL, United States; provided gratis from the company), a product actively marketed for usage in cigarette filters.²² To compare to previous literature,⁷ Anasorb CSC charcoal (SKC Inc., Eighty Four, PA, United States) was purchased.

Figure 1.

In-line charcoal filtration setup with the smoke machine. (A) Photograph of the entire setup with cigarette and impinger. (B) Zoom-in of setup to show the direction of smoke, going from 1 to 4, including an inset of the holder with the packed charcoal (200 mg) and glass wool holding the charcoal in place. Zero milligram charcoal experiments were performed with the same setup with glass wool but without charcoal.

Derivatization of Carbonyls.

DNPH solution was made as described previously^15,17,23 by dissolving 1 g recrystallized DNPH in 50 mL of diglyme and 150 mL of ACN with 360 μL of 12 N HCl. Similar to previous work,²³ the mainstream smoke generated from one cigarette by the CSM-HPP was pumped through Tygon tubing to the charcoal holder and then into an impinger containing 10 mL of DNPH solution. After smoking, the solution was transferred to a 15 mL vial; 500 μL of pyridine was added, and then the vial was stored at 4 °C until HPLC-UV analysis. We performed three replicates for all conditions unless otherwise mentioned. Although samples were found to be stable for at least 2 weeks at 4 °C, all HPLC-UV analyses were performed within 3 days of collection to permit time for any needed reanalysis. Peak confirmation was performed via GC-MS for all carbonyls using the 3R4F on the HCI method without charcoal (Supplemental Figures 1−7).

High Performance Liquid Chromatography with Ultraviolet Detection (HPLC-UV) Analysis.

HPLC-UV analyses were performed with a binary system containing two Waters (Milford, MA, United States) 510 pumps, a Shimadzu (Kyoto, Japan) SPD-10A VP UV–vis detector, and a Hitachi (Tokyo, Japan) D-2500 Integrator. The method used was based on the CORESTA method but optimized for use with two pumps.²⁴ This method has been described previously²³ with >98% recovery and approximately 12% precision for all carbonyls. Briefly, a C18 column (Bondclone, 10 μm × 300 mm × 3.9 mm; Phenomenex, Torrance, CA, United States) separated the carbonyls using two mobile phases: (A) 59% water, 30% acetonitrile, 10% tetrahydrofuran, and 1% isopropanol and (B) 65% acetonitrile, 33% water, 1% tetrahydrofuran, and 1% isopropanol. The elution gradient was: 0 min, 100% A; 8 min, 70% A; 16 min, 60% A; 20 min, 54% A; 22 min, 40% A; 25 min, 100% A; and 31 min. 100% A. All measurements were carried out at room temperature (22 ± 1 °C). The flow rate was 1.5 mL/min. The detection wavelength was 365 nm. All sample injections (20 μL) were injected by a Hewlett-Packard (Palo Alto, CA, United States) Series 1050 autosampler.

Nicotine and Total Particulate Matter (TPM) Determination.

Total nicotine and TPM from mainstream smoke was trapped onto a Cambridge filter pad (Performance Systematix Inc., Grand Rapids, MI, United States) placed in-line after the charcoal holder. The nicotine was then extracted and analyzed by GC-FID as previously described.²⁵ TPM was determined by weighing the Cambridge filter before and after smoking on an analytical balance.

Measurement of Charcoal in Commercial Cigarette Filters.

Filters were separated into charcoal and noncharcoal containing segments. The remaining fibers of cellulose acetate were dissolved by addition of 3 mL of acetone. After the charcoal was allowed to settle, the acetone was removed, and the remaining charcoal was allowed to dry for 3 h in a fume hood prior to reweighing. Three replicates were performed for each brand. The method showed complete recovery when 200 mg of charcoal was tested through the same procedure.

Statistical Analyses.

For all carbonyl comparisons, one-way ANOVAs with Tukey contrasts were used to evaluate all pairwise comparisons presented as the data appeared to meet ANOVA assumptions. t tests were performed for comparing nicotine and TPM results. All statistical analyses were generated using SAS software Version 9.4 of the SAS System (SAS Institute Inc., Cary, NC, United States).

RESULTS

Commercial Charcoal-Filtered Cigarettes.

Initially, we compared carbonyl deliveries from three available commercial charcoal-filtered cigarette brands with those for a reference cigarette (3R4F), which has been shown^16,17 to be representative of carbonyl production in many popular United States brands (Figure 2). Under the less intensive ISO protocol, carbonyl yields from two of the three charcoal cigarettes (Lark and Parliament) were significantly lower than the third charcoal cigarette (Kent) and the 3R4F cigarette. However, no significant differences were observed using the more intensive HCI method. Examination of the filters (Figure 2) revealed low levels of charcoal incorporated into the filter using an “on-tow” design (i.e., filters with charcoal speckled over cellulose acetate). The weight of charcoal in the filters ranged from 40 to 70 mg for each cigarette brand (Parliament: 46.9 ± 2.4 mg; Kent: 70.0 ± 1.6 mg; Lark: 61.5 ± 5.4 mg). In addition, both Parliament and Lark had a distinct ring of clustered ventilation holes in the filter, indicating that these brands may be more highly ventilated. When examining tobacco content, both Parliament and Lark (0.62 ± 0.02 and 0.52 ± 0.02 g, respectively) contained less tobacco than Kent (0.69 ± 0.02 g) and 3R4F (0.78 ± 0.02 g). The number of puffs for each cigarette on the ISO method (Kent: 7.5 ± 0.7; Lark: 7.0 ± 0.0; Parliament: 7.0 ± 0.0; 3R4F: 9.0 ± 0.0) and HCI method (Kent: 8.5 ± 0.7; Lark: 6.0 ± 0.0; Parliament: 7.5 ± 0.7; 3R4F: 11.0 ± 0.0) were strongly correlated with tobacco amount (ISO: r² = 0.78; HCI: r² = 0.97); thus, similar trends in carbonyl delivery between products were observed when data were expressed per gram of tobacco (data not shown).

Figure 2.

Carbonyl deliveries (n = 2–4) from commercial charcoal cigarettes and using both the ISO and HCI methods. All bars are means with standard deviation with symbols for statistical significance. The photos of the filters of each commercial charcoal cigarette are also shown.

Make-Your-Own Cigarettes.

As this work is one of the first to examine the carbonyl delivery from MYO cigarettes, we first analyzed and compared levels of carbonyls in mainstream smoke from noncharcoal-filtered MYO cigarettes made with either pipe tobacco (Good Stuff Red) or cigarette tobacco (Rave Red) with those from 3R4F cigarettes (Table 1; Figure 3A). While similar results were obtained between cigarettes for most carbonyls, several differences, particularly for the MYO made with Good Stuff using the ISO method, were apparent. Good Stuff tended to deliver lower levels of acrolein and crotonaldehyde and higher levels of propionaldehyde than 3R4F, whereas Rave only differed from 3R4F for propionaldehyde (delivering higher levels) under the ISO protocol (P < 0.05). On the HCI method, the only significant difference observed was between Good Stuff and 3R4F for acetaldehyde delivery (P < 0.05).

Table 1.

Carbonyl Delivery for MYO Cigarettes and the 3R4F Research Cigarette with Different Amounts of Charcoal on Both the ISO and HCI Methods^a

brand	cigarette type	charcoal amount mg	formaldehyde μg/puff	acetaldehyde μg/puff	acetone μg/puff	acrolein μg/puff	propionaldehyde μg/puff	crotonaldehyde μg/puff	methyl ethyl ketone μg/puff	puff number
ISO method
Good Stuff Red	MYO	0	2.18 ± 0.11a	54.14 ± 4.47a	19.09 ± 0.36a	2.99 ± 0.15a	4.91 ± 0.38a	0.49 ± 0.24a	2.14 ± 0.14a	9 ± 1a
		50	1.83 ± 0.14b	22.56 ± 6.36b	5.45 ± 2.23b	1.69 ± 0.66b	0.74 ± 0.48b	0.25 ± 0.08a	0.18 ± 0.18b	9 ± 0a
			16%	58%	71%	43%	85%	49%	92%
		200	1.51 ± 0.05c	5.66 ± 2.41c	1.54 ± 0.74c	0.53 ± 0.31c	0.39 ± 0.03b	0.18 ± 0.04a	N.D.b	9 ± 0a
			31%	90%	92%	82%	92%	63%	~100%
Rave Red	MYO	0	2.22 ± 0.19a	59.05 ± 4.82a	18.78 ± 2.66a	4.90 ± 0.12a	4.84 ± 0.78a	0.83 ± 0.17a	3.30 ± 0.52a	9 ± 0a
		50	0.53 ± 0.04b	22.73 ± 3.96b	4.46 ± 0.61b	1.19 ± 0.20b	1.07 ± 0.12b	1.29 ± 0.10b	0.75 ± 0.07b	9 ± 0a
			76%	62%	76%	76%	78%		77%
		200	0.12±0.15c	3.81 ± 1.56c	0.91 ± 0.57b	0.25 ± 0.13c	0.43 ± 0.32b	N.D.c	0.23 ± 0.11b	9 ± 0a
			95%	94%	95%	95%	91%	~100%	93%
3R4F	Premade	0	1.96 ± 0.03a	50.39 ± 2.76a	17.99 ± 1.05a	5.44 ± 0.47a	3.22 ± 0.00a	1.35 ± 0.02a	2.59 ± 0.05a	8 ± 1a
		50	1.12 ± 0.08b	24.75 ± 2.25b	5.87 ± 0.59b	1.62 ± 0.20b	0.73 ± 0.13b	1.29 ± 0.05a	0.86 ± 0.02b	8 ± 1a
			43%	51%	67%	70%	77%	4%	67%
		200	0.46 ± 0.11c	3.52 ± 0.98c	0.59 ± 0.23c	0.25 ± 0.18c	NDc	NDb	0.19 ± 0.04c	8 ± 1a
			77%	93%	97%	95%	~100%	~100%	93%
HCI method
Good Stuff Red	MYO	0	3.80 ± 0.69a	100.01 ± 9.69a	41.08 ± 8.36a	11.60 ± 1.86a	7.96 ± 1.63a	2.99 ± 0.72a	8.20 ± 1.98a	11 ± 1a
		50	3.98 ± 1.24a	64.43 ± 2.27b	21.39 ± 1.73b	5.55 ± 0.43b	4.18 ± 0.07a,b	2.46 ± 0.20a	5.27 ± 1.17a,b	11 ± 0a
				36%	48%	52%	47%	18%	38%
		200	2.67 ± 0.53a	28.09 ± 8.03c	7.76 ± 3.14b	2.33 ± 0.88b	2.05 ± 1.08b	0.35 ± 0.13b	1.79 ± 0.63b	11 ± 0a
			30%	72%	81%	80%	74%	88%	78%
Rave Red	MYO	0	4.61 ± 1.16a	115.70 ±2.10a	46.24 ± 4.85a	12.92 ± 1.03a	8.86 ± 0.84a	4.80 ± 1.07a	9.62 ± 1.45a	10 ± 1a
		50	2.51 ± 0.42b	70.02 ± 3.45b	19.10 ± 2.02b	5.04 ± 0.58b	4.29 ± 0.51b	1.29 ± 0.19b	3.24 ± 0.57b	10 ± 1a
			46%	39%	59%	61%	52%	73%	66%
		200	1.81 ± 0.71b	25.45 ± 14.34c	4.99 ± 3.21c	1.41 ± 0.98c	1.16 ± 0.62c	1.06 ± 0.21b	0.82 ± 0.37c	10 ± 1a
			61%	78%	89%	89%	87%	78%	91%
3R4F	Premade	0	5.39 ± 0.18a	131.05 ± 11.82a	50.63 ± 5.16a	15.62 ± 1.19a	8.88 ± 0.49a	4.22 ± 0.08a	10.53 ± 3.38a	11 ± 0a
		50	2.77 ± 0.44b	86.59 ± 8.23b	25.36 ± 2.05b	6.19 ± 0.23b	4.68 ± 0.32b	1.22 ± 0.42b	4.56 ± 0.35b	11 ± 1a
			49%	34%	50%	60%	47%	71%	57%
		200	1.85 ± 0.15c	31.11 ± 12.43c	7.86 ± 3.83c	1.74 ± 0.98c	1.53 ± 0.68c	0.47 ± 0.22b	1.68 ± 0.66b	10 ± 1a
			66%	76%	84%	89%	83%	89%	84%

^aValues are mean ± standard deviation with percent reduction relative to 0 mg charcoal. Different superscripts (a, b, or c) within a carbonyl group across charcoal amounts denote significant differences as determined by ANOVA (p < 0.05).

Figure 3.

(A) Carbonyl deliveries from MYO cigarettes made with pipe (Good Stuff, white) and cigarette (Rave, light gray) tobacco using both the ISO and HCI methods. The 3R4F research cigarette (dark gray) is used for comparison as the carbonyl delivery from the 3R4F has been shown to be representative of delivery from commercial cigarettes.^16,17 (B) MYO cigarettes made with pipe tobacco (Good Stuff) with Cartel Cellulose Acetate Filter (white) and Cartel Charcoal Filter Tubes (gray, image under graph) using both the ISO and HCI methods. All bars are means with standard deviation with symbols for statistical significance. The photo of the filter of the MYO charcoal cigarette tube is also shown.

We then tested the effects of charcoal in the cigarette filters by comparison of the carbonyl delivery of MYO cigarettes made with Good Stuff tobacco and filters containing cellulose acetate and charcoal or cellulose acetate alone (Figure 3B). The charcoal amount in the MYO filters was 34.3 ± 1.7 mg. Results showed that the charcoal filters included in MYO tubes were ineffective at reducing the majority of carbonyl levels on the ISO method. There were no significant differences in any of the target carbonyls between the charcoal and noncharcoal MYO cigarettes on the HCI method. Upon examination, the MYO charcoal filter showed similar charcoal amounts as those in the commercial charcoal cigarettes.

To determine the impact of higher quantities of charcoal configured in a plug-like design, we utilized an in-line charcoal filtration for both MYO and 3R4F cigarettes (Figure 1). For all carbonyls, a dose-responsive trend of decreasing levels with increasing charcoal amount (0–200 mg) was observed for all cigarettes tested under both ISO and HCI protocols (Figure 4; Supplemental Figure 8) with no change in the number of puffs (Table 1). The levels of reduction were similar between both MYO cigarettes and 3R4F premade cigarettes. Overall reductions were greater using the ISO methods compared to the HCI method (Table 1). The greatest reductions tended to occur for acetaldehyde, acetone, propionaldehyde, and methyl ethyl ketone (90–100% at the highest charcoal dose for ISO and 72–91% at the highest dose for HCI). Formaldehyde was less affected by charcoal than other carbonyls (reduced by 31–95 and 30–66% at the highest charcoal dose on the ISO and HCI methods, respectively).

Figure 4.

Carbonyl delivery for MYO cigarettes and the 3R4F research cigarette with different amounts of charcoal on both the (A) ISO and (B) HCI methods. Bars are mean ± standard deviation.

To further investigate if this effect was specific to the charcoal we used, we also tested the 3R4F with the charcoal used in the previous work by the CDC.⁷ While from a similar source (coconut shell), this charcoal was slightly different in mesh size (20/40 [SKC] vs 30/70 [Donau]). Results indicate that at the same amount of charcoal, similar reductions were observed (Figure 5) for most carbonyls.

Figure 5.

Delivery of each carbonyl (μg/puff) from the 3R4F research cigarette on the HCI method, comparing no charcoal (white) to both Donau charcoal (solid) and SKC Inc. charcoal (dotted) at 50 mg (light gray) and 200 mg (dark gray). All bars are means with standard deviation.

Nicotine and TPM.

We also tested the effect of charcoal on the delivery of other tobacco smoke constituents, nicotine and TPM, in mainstream smoke from 3R4F cigarettes (Figure 6). The delivery of both nicotine and TPM were similarly reduced (28 and 20%, respectively) by filtration with 200 mg of charcoal regardless of smoking protocol (ISO or HCI). However, these reductions were only statistically significant (P < 0.05) using the HCI method.

Figure 6.

Percent recovery of nicotine and TPM at 200 mg charcoal for the 3R4F cigarette on the ISO and HCI methods. All bars are means with standard deviation. Statistically significance (p < 0.05) differences in delivery were observed only on the HCI method as noted by asterisks.

DISCUSSION

Overall, our findings suggest that charcoal filtration can be an effective means of reducing carbonyl levels in mainstream cigarette smoke, so long as an adequate amount (e.g., 200 mg) of charcoal is used in a plug-type configuration. These findings are consistent with other previously reported results on carbonyls and their reported levels of reduction with plug-type charcoal filtration.^5,7,8 We also found that the delivery of most carbonyls tended to be lower in some commercially available charcoal-filtered cigarette brands compared to noncarbon filtered cigarettes under nonintensive smoking conditions (ISO), while no differences were apparent under intensive smoking conditions (HCI). These findings, which are consistent with previous reports,^5,6 suggest that the levels of charcoal in commercial brands with the in-tow design may be insufficient for measurable carbonyl reduction under intensive smoking conditions. However, to rule out other possible differences between charcoal and noncharcoal brands, such as tobacco blend, we also compared the effectiveness of commercially available in-tow filters using the same tobacco in MYO cigarettes. Similar results were obtained with minimal reductions being observed for some carbonyls under ISO conditions, but no differences found under HCI conditions. These findings support the conclusion that many of the commercially available charcoal filters are minimally effective based on their design and low levels of charcoal.

In a previous study, Morabito et al.⁷ modified existing filters in commercial cigarettes by adding a “plug” of charcoal and observed significant reductions in carbonyl delivery. However, the modification, which included cutting and taping the filter, could potentially introduce artifactual changes in carbonyl delivery and also prohibit the use of ISO methods due to filter blocking to keep the charcoal in place. Thus, to specifically determine the impact of charcoal, we used a plug-style charcoal filter placed in-line downstream of the cigarette, which obviated the need for modifying the existing cigarette filter. Similar to the previous finding, we observed substantial dose-responsive reductions in carbonyls with charcoal under HCI conditions for both conventional and MYO cigarettes. We also observed even greater reductions in carbonyls under ISO conditions.

Overall, the level of reduction depended on the specific carbonyl, type of cigarette, and level of charcoal, with greatest reductions occurring for 200 mg of charcoal. In general, the greatest reductions were observed for crotonaldehyde, methyl ethyl ketone, propionaldehyde, acrolein, and acetone (74–91% under HCI and 82–100% under ISO). Reductions were also observed for formaldehyde (30–66% under HCI and 31–95% under ISO) and acetaldehyde (90–94% under HCI under 82– 100% for ISO). As all of these carbonyls are associated with negative disease outcomes,² such substantial reductions in their delivery from cigarette smoke could potentially lead to reductions in harm for smokers; however, more work is needed to confirm this. Previous literature has shown that charcoal can be “poisoned” over the time course of smoking the cigarette, decreasing its efficiency at filtering carbonyls out of cigarette smoke.^26,27 While we did not test each puff individually, the high efficiency of 200 mg of charcoal at reducing carbonyls suggests that there is little breakthrough of carbonyls and no indication that charcoal poisoning might be occurring. However, it is possible that there is a greater risk for breakthrough when using the more intensive HCI method as it allows for higher overall levels of exposure to smoke.

In the dose–response experiments, significant reductions in carbonyls were observed with 50 mg of charcoal. This is in contrast to results with commercially available charcoal-filtered cigarettes which contained comparable amounts of charcoal (40–70 mg) but appeared less effective at eliminating carbonyls. This is likely due to differences in filter design with the plug-style being more effective, potentially allowing for greater contact of tobacco smoke carbonyls with charcoal than with the on-tow design. Certainly, with the latter design, where charcoal is dispersed unevenly throughout the filter, there is a high likelihood that some of the smoke can pass through without interacting with charcoal (Figure 2). Thus, based on these and previous findings,^5,7 the optimal charcoal filter design for minimizing carbonyl delivery is a plug-style containing ≥200 mg of charcoal.

In all experiments and cigarettes tested, the level of carbonyl reduction was substantially greater under ISO conditions compared to HCI conditions. These differences likely reflect differences in smoking intensity (flow rate) as the HCI method was developed more recently to reflect the more intensive puffing of modern smokers and a “maximum” limit for product comparisons,²⁸ as the exposure of most smokers falls between the ISO and HCI methods.²⁹ We speculate that under the less intensive ISO method, there is greater time for interaction between carbonyls in mainstream smoke and the charcoal particles in the stationary phase as a result of a reduced flow rate. The actual reduction in an individual smoker will depend on their specific smoking behavior (topography), which is subject to a high degree of interindividual variation.³⁰ Thus, the actual effectiveness of charcoal filtered cigarettes would be expected to vary considerably among different smokers.

Of the two different types of tobacco used in the MYO experiments (cigarette or pipe tobacco), similar results were obtained relating to carbonyl delivery and its reduction by charcoal filtration. However, these represent only two of many brands that are available on the market. As we have shown previously that there are wide variations on the levels of carbonyl delivery among popular cigarette brands,¹⁷ the same may be true for MYO tobaccos.

While charcoal reduction may represent a feasible method for reducing the delivery of toxic gas phase components such as carbonyls from mainstream tobacco smoke, the effectiveness of this approach as a potential risk reduction strategy will also depend upon the impact of the charcoal on nicotine, the primary addictive agent in tobacco smoke, and on other toxic agents found in mainstream tobacco smoke. As observed previously,⁷ we found that at the highest dose of charcoal used, nicotine was only reduced by <30%, regardless of smoking protocol, and only significantly reduced on the more intense protocol. This is potentially an important factor regarding the feasibility of this approach, because if there were large reductions in nicotine, addicted smokers would either choose not to smoke these products or could potentially smoke more to makeup for the reduction in nicotine delivery. Also, consistent with previous reports,⁷ we found that charcoal filtration had minimal effects on total particulate matter (~20% reduction). Because this phase of mainstream smoke contains many known toxic and carcinogenic agents, including polycyclic aromatic hydrocarbons and nitrosamines, charcoal is likely to have little effect on harm related to these exposures. Overall, to fully understand the potential benefit of charcoal as a potential harm-reduction strategy, a better understanding of the relative impact of carbonyl reduction on toxicity will be needed; however, the demonstrated removal of carbonyls in cigarette smoke is a beneficial starting point.