October 22, 2006

 

David W. Kuneman

Director of Research

The Smoker’s Club, Inc.

 

There have been recent reports in the media,

READ

and at least one published scientific paper claiming casino workers in Nevada have genetic damage caused by secondhand smoke which was published in Toxicology Letters July, 2005, by Pritsos, et.al. of the University of Nevada’s Department of Nutrition.

READ

  

In an interview, Pritsos stated a more exhaustive analysis of the relationship between secondhand smoke exposure and DNA damage in casino workers would be published later this year in The Journal of the American Medical Association, but as of October, 19, 2006, this has not occurred.

 

The authors of this July 2005 journal article actually concluded nonsmoking female workers exposed to higher levels of secondhand tobacco smoke were not more likely to have genetic damage, but male nonsmoking workers exposed to more secondhand smoke had more 8OHdG and general gene damage. They used serum continine levels to assess secondhand smoke exposure; continine is a metabolite of nicotine and is considered tobacco-smoke-specific biomarker. The original abstract is reproduced below.

 

 

Abstract

There is much data implicating environmental tobacco smoke (ETS) in the development and progression of disease, notably cancer, yet the mechanisms for this remain unclear. As ETS is both a pro-oxidant stressor and carcinogen, we investigated the relationship of ETS exposure to intracellular and serum levels of DNA-damage, both oxidative 8-hydroxy-2-deoxyguanosine (8OHdG) and general, in non-smokers from non-smoking households, occupationally exposed to ETS. General DNA-damage consisting of single and double strand breaks, alkali-labile sites and incomplete base-excision repair, increased significantly in a dose-dependent manner with ETS exposure in men (P = 0.015, n = 32, Pearson) but not women (P = 0.736, n = 17).

 

Intracellular 8OHdG-DNA-damage and general DNA-damage were both greater in men than women (P = 0.0005 and 0.016, respectively) but 8OHdG serum levels did not differ between the genders. Neither 8OHdG-DNA-damage nor serum levels correlated with increasing ETS exposure. This is the first study to demonstrate dose-dependent increases in DNA-damage from workplace ETS exposure. Perhaps most interesting was that despite equivalent ETS exposure, significantly greater DNA-damage occurred in men than women. These data may begin to provide a mechanistic rationale for the generally higher incidence of some diseases in males due to tobacco smoke and/or other genotoxic stressors.

© 2005 Elsevier Ireland Ltd. All rights reserved.

 

The article found male 8OHdG damage was higher than females, but not dose-dependant. Therefore the authors correctly concluded since this kind of damage is not dose-dependant, that this damage cannot be associated with secondhand smoke exposure.  However, they also concluded that general genetic damage is dose-dependant in male casino workers. In summary, of the two kinds of genetic damage they could have found among both sexes, only one kind was actually found in male nonsmokers. This is one-out-of-four possibilities. 

 

Following the logic these authors used with respect to dose-dependency, and 8OHdG damage, it becomes imperative to determine if general gene damage in males actually is dose-dependant with respect to secondhand smoke exposure. The actual graphs from the published paper are reproduced below. Tabulated data used to prepare these graphs were not presented in the published paper.

As the authors correctly concluded, an simple examination of the graph for female casino workers shows us that there is no genetic damage increase with increasing secondhand smoke exposure. However, a simple examination of the graph for males also does not show increasing genetic damage with increasing secondhand smoke exposure. I agree with the authors these male data do not show a dose-dependant relationship for 8OHdG damage, I disagree with the authors’ claim that these data do show a dose-dependant relationship between secondhand smoke exposure and general gene damage. Females are not that genetically different than males, in fact, most secondhand smoke studies claim nonsmoking females are more susceptible than nonsmoking males. Because these authors readily admit they did not find general gene damage in females with increasing secondhand smoke exposure, they should have been suspicious of their conclusions regarding general gene damage in males.

 

As stated, the authors do not provide the raw data used to prepare the graphs. However, I was able to enlarge the graph for males and measure the triangulated data points, representing general genetic damage. I then prepared the following table of the data for male general genetic damage that the authors must have used for the 32 males included in the study. I believe these data are very close to the author’s data.

 

Continine conc	DNA damage
4.9	.01
3.0	.16
2.4	.92
2.2	.01
2.1	0
2.1	0
1.9	.13
1.6	.21
1.4	.21
1.4	1.22
1.4	0
1.4	.7
1.3	0
1.2	0
.9	.55
.4	.72
.9	.1
.8	.03
.5	.12
.4	.1
.3	.3
.3	.08
.7	.39
.8	.39
.6	.01
.6	.01
.4	0
.2	0
.2	.03
.5	0
.5	0
.5	0

 

 

At a glance, it is easily concluded there are just as many high secondhand smoke exposure levels associated with low general gene damage, as there are low exposure levels associated with high general gene damage.  This is one of the main faults of this paper.  Note the highest damage score (1.22) had a continine level of 1.4 which was closer to zero continine, than the highest continine level found in this study, which was 4.9. The worker with 4.9 continine concentration had practically no general gene damage. The authors provided no data on the past smoking status of these male workers either.( Appendix A)  It is possible this male with a score of 1.22 is a former heavy smoker. The only exposure data provided is continine levels which reflect recent exposure. Genetic damage takes years to develop. The possibility some of these workers are former smokers could explain why some with lower recent exposure had more damage. It is also possible the highest exposure level, (4.9) had  a low genetic damage score because he is a recent new hire, and perhaps never smoked or was previously employed in  a smoke-free environment. It is particularly bothersome that all these males and females work in casinos, but the exposure levels vary from practically zero, to as much continine as is found in occasional smokers. It is highly implausible that the secondhand smoke concentrations in casino air vary this widely and therefore the proper conclusion is that continine measurements are actually not good indicators of secondhand smoke exposure in casinos.  

 

Attempts were made to find a regression which would prove these male nonsmoking casino workers’ data make sense. (Appendix B) The use of a nonlinear polynomial regression test is the kind of test most likely to find a statistical relationship, if one does indeed exist. In fact, absolutely no statistical relationship exists. These data actually predict a little secondhand smoke exposure is harmful to male casino workers in terms of general DNA damage, but that more secondhand smoke exposure has a preventative effect. The “best fit” least squares polynomial regression developed from the nonsmoking male workers’ data predicts general genetic damage increases up to  approximately 1.2 continine score, then genetic damage decreases to approximately zero at a continine score of 5.0. Contrary to the authors’ attempts to conclude secondhand smoke causes general genetic damage in nonsmoking male workers because a dose-response relationship exists, in fact, an inverse relationship exists for much of the data representing higher doses. 

 

 The authors went to considerable effort to provide data on age, body mass index, ethnicity, marital status, and education, but did not provide any data on ever-smoking status, length of casino employment, other employment history with respect to possible genetic risk, former smoking status of spouse, alcohol use, diet, etc. These could have more impact on each individual’s genetic score than casino smoke. For example, former smoking status imparts a risk for lung cancer three times that of a never smoker. The literature consensus is that secondhand smoke exposure in the workplace imparts a risk 1.3 times the risk of a never smoker.  If all these consensus claims are true, then former smoking status is seven times more risky  than secondhand smoke exposure. In the absence of controlling for these other confounding variables, it is impossible for the authors to have concluded general genetic damage in male nonsmoking casino workers is caused by secondhand smoke.

 

What could have helped the authors draw more solid conclusions, would have been to also include nonsmoking workers who have nonsmoking spouses and are employed in nonsmoking workplaces. If their genetic scores, and 8OHdG data were similar to those of the casino workers, then this study would be demonstrably inconclusive. Careful researchers indeed would have included a control group. However, researchers attempting to influence the Nevada voters’ outcome on the propositions to ban public place smoking, would be more likely to omit these very important considerations from their paper.

 

* Appendix A:  Confounding variables considered By Pritsos et. al.

 

 

Appendix B:  Attempt to find a statistical relationship between general genetic damage and serum continine concentration in male nonsmoking casino workers:

Instructions on XURU.org website I used to test for statistical significance : This page allows you to work out polynomial regressions, also known as polynomial least squares fittings. For the relation between two variables, it finds the polynomial function that properly fits a given set of data points. The result is not necessarily the best possible, but usually it is a very good one and further improvement possibilities are small. In the case that the selected degree is equal to the number of data points less one a polynomial interpolation results

 

• Insert manually & See details: If you prefer you can insert all the points manually, for which you first have to specify the number of data points. You also can see details of the calculation -as the calculated value of y and the error at each point- in this area. Enter the number of data points: Select polynomial degree:
	x		y			Calculated y		Error
1.	4.9	.01				-1.726075881·10-2		2.726075881·10-2
2.	3.0	.16				0.270793832		0.110793832
3.	2.4	.92				2.944710029·10-1		6.255289971·10-1
4.	2.2	.01				2.951860667·10-1		2.851860667·10-1
5.	2.1	0				2.941978498·10-1		2.941978498·10-1
6.	2.1	0				2.941978498·10-1		2.941978498·10-1
7.	1.9	.13				2.895299186·10-1		1.595299186·10-1
8.	1.6	.21				2.757992782·10-1		6.579927817·10-2
9.	1.4	.21				2.621596888·10-1		5.215968875·10-2
10.	1.4	1.22				2.621596888·10-1		9.578403112·10-1
11.	1.4	0				2.621596888·10-1		2.621596888·10-1
12.	1.3	0				2.621596888·10-1		4.378403112·10-1
13.	1.4	.7				2.539941453·10-1		2.539941453·10-1
14.	1.2	0				0.244931436		0.244931436
15.	.9	.55				2.123603133·10-1		3.376396867·10-1
16.	.4	.72				1.401317923·10-1		5.798682077·10-1
17.	.9	.1				2.123603133·10-1		1.123603133·10-1
18.