Back

A REVIEW OF THE EPIDEMIOLOGY

OF LUNG CANCER AND

TYPE OF CIGARETTE SMOKED

Peter N Lee

P.N.Lee Statistics and Computing Ltd

17 Cedar Road

Sutton

Surrey, SM2 5DA

UK

June 2000

EXECUTIVE SUMMARY

This review focuses on 54 studies which have presented results relating type of cigarette smoked to risk of lung cancer, 17 conducted in the USA (one partly in Canada), 10 in the UK, 12 in the rest of Europe, nine in Asia and six in South and Central America. Thirteen of the studies were prospective, 39 case-control and two of other designs.

Difficulties in assessing the relative risk of lung cancer associated with smoking different types of cigarette from the available evidence include the following:

• Inaccuracy in classifying subjects into the groups being compared;
• Inadequate adjustment for potential confounding variables. Of the 54 studies reviewed, 15 did not adjust for any potential confounding factors at all, only 34 adjusted for age, 29 for number of cigarettes smoked, 12 for duration of smoking and 17 for any factor other than age or smoking;
• Difficulties in obtaining results from different studies in a comparable format;
• Limited nature of the evidence on some aspects of type of cigarette smoked. Despite the continuing decline in tar levels of cigarettes smoked, there is virtually no evidence at all in relation to deaths or cases of lung cancer occurring in the 1990s;
• Inadequate reporting of some major studies. Of the two largest case-control studies, one (the West European study conducted in five countries) failed to adjust for age in any of its analyses, while the other (American Health Foundation Multicentre Study II) failed to adjust for any variable in its analyses of tar level. Of the two largest prospective studies, one (CPSI) never presented any results for filter/plain smoking, while the other (CPSII) has, for males, never presented any results by tar level of brand smoked and has only presented results for filter/plain smoking graphically, without any indication of variability;
• Many of the analyses only relate to smoking specific types of cigarette for a limited period of time and do not provide direct information on lifetime smoking of that type of cigarette.

Despite these difficulties a number of consistent findings emerge from the review.

Filter/plain (36 studies) Over half of the studies provide statistically significant evidence of a lower risk of lung cancer in smokers of filter cigarettes. Meta-analyses of relative risks for filter/plain (comparing the most extreme groups where results were presented in more than two categories) demonstrated a highly significant (p < 0.001) advantage to filter (RR = 0.58, CI = 0.55-0.62 for males, n = 28 and RR = 0.67, CI = 0.59-0.75 for females, n = 14). The advantage was evident separately in studies conducted in the USA, in the UK, in Europe, and in Asia and South America. Although studies varied in respect of confounding variables taken into account, there was no evidence that this had an important effect on the conclusions.

Eight studies provided evidence on filter/plain relative risk by histological type. For squamous cell carcinoma (or Kreyberg I), all the sex-specific estimates were below 1, with the combined estimate (RR = 0.56, CI = 0.50-0.62, n = 11) showing a highly significant (p < 0.001) advantage to filter cigarettes. For adenocarcinoma (or Kreyberg II) the advantage was less marked (RR = 0.84, CI = 0.70-1.00, n = 8). However, there was no indication of an increase in risk associated with the switch to filter cigarettes, contrary to indirect inferences drawn from the observed rise in incidence of adenocarcinoma in the USA and the assumed greater distribution of smoke components distally in the lung when filter cigarettes are smoked (Thun et al, 1997a).

Tar level (16 studies) The results show a clear pattern of decreasing risk with declining tar levels. For males, meta-analysis of results comparing the lowest and highest tar categories gave a highly significant (p < 0.001) lower risk in the low tar category (RR = 0.77, CI = 0.69-0.86, n = 13). For females, a similar advantage was seen (RR = 0.82, CI = 0.70-0.97, n = 9). The advantage to filter cigarettes is equivalent to a 2-3% reduction in risk for each mg tar per cigarette. Limited evidence suggests that these conclusions were not materially affected by failure to adjust for potential confounding variables in some studies. Only one study reported results by histological type. That study reported a lower risk of Kreyberg I lung cancer in lower tar smokers in females, but no effect of tar on risk of Kreyberg I lung cancer in males or in Kreyberg II lung cancer in either sex.

Hand rolled/manufactured (12 studies) Compared to smokers of manufactured cigarettes only, a significantly higher risk of lung cancer was seen in smokers who had ever smoked hand rolled cigarettes (RR = 1.41, CI = 1.26-1.57, n = 15), in smokers of hand rolled cigarettes only (RR = 1.27, CI = 1.09-1.48, n = 12) and in mixed smokers of hand rolled and manufactured cigarette smokers (RR = 1.26, CI = 1.06-1.49, n = 11). Adjustment for potential confounding variables was generally quite poor in these studies. Limited evidence did not clarify the relevance of cell type in this comparison.

Black (dark)/blond (light) (10 studies) Compared to smokers who have never smoked black cigarettes, a remarkably consistent higher risk of lung cancer was seen in those who have ever smoked black cigarettes (RR = 1.71, CI = 1.50-1.96, n = 12). The increase was evident both for smokers of black cigarettes only (RR = 1.72, CI = 1.44-2.05, n = 8) and for mixed smokers of black and blond cigarettes (RR = 1.47, CI = 1.21-1.79, n = 8). The increase in lung cancer for ever smokers of black cigarettes was evident both for squamous cell carcinoma (RR = 1.96, CI = 1.44-2.67, n = 4) and for adenocarcinoma (RR = 1.64, CI = 1.17-2.32, n = 4).

Mentholated cigarettes (3 studies) The combined evidence is difficult to interpret, showing a non-significantly increased risk in smokers of mentholated cigarettes in men (RR = 1.18, CI = 0.91-1.53) but a significantly decreased risk in women (RR = 0.70, CI = 0.52-0.95).

Other cigarette types The report also presents limited evidence relating lung cancer risk to the nicotine level of the brand smoked, to the smoking of bidis, to the smoking of brands local to Okinawa and to the smoking of pillis and pöllis, but no clear conclusions could be drawn.

Note All the combined relative risk estimates cited above are based on fixed-effects meta-analysis. Estimates from random-effects meta-analyses, also shown in the main body of the report, were generally quite similar.

INDEX

Text Page

1. Introduction 1

2. Methods 3

3. Study characteristics 6

4. Some general problems in studying the relationship of lung cancer to type 12 of cigarette smoked

5. Lung cancer risk in filter and plain cigarette smokers 18

6. Lung cancer risk by tar level of cigarette smoked 25

7. Lung cancer risk in hand rolled vs. manufactured cigarette smokers 29

8. Lung cancer risk in black (dark) vs. blond (light) cigarette smokers 32

9. Lung cancer risk by other aspects of type of cigarette smoked 35

10. Discussion and conclusions 40

11. Acknowledgements 46

12. References 47

Tables

3.1 Studies providing information on risk of lung cancer in relation to type of cigarette smoked T1

3.2 Number of studies including lung cancer cases or deaths in specified periods T4

3.3 Lung cancer cases in the 54 studies T5

3.4 Controls (or populations at risk) in the 54 studies T8

3.5 Aspects of cigarette type considered T11

3.6 Potential confounding variables adjusted for T14

5.1 Details of studies providing evidence on risk of lung cancer in filter and plain cigarette smokers T17

5.2 Relative risk (95% CI) of lung cancer in filter and plain cigarette smokers T21

5.3 Relative risk (95% CI) of lung cancer in relation to filter and plain cigarette smoking (using the most extreme groups for comparison where more than two groups were compared) T29

5.4 Relative risk (95% CI) of lung cancer in filter and plain cigarette smokers - by histological type T32

5.5 Relative risk (95% CI) of squamous cell carcinoma (or Kreyberg I) and of adenocarcinoma (or Kreyberg II) in relation to filter and plain cigarette smoking (using the most extreme groups for comparison where more than two groups were compared) T35

5.6 Effect of adjustment for various risk factors on relative risk (95% CI) of lung cancer in relation to filter and plain cigarette smoking T36

6.1 Details of studies providing evidence on risk of lung cancer in relation to tar level T38

6.2 Relative risk (95% CI) of lung cancer in relation to tar yield and brand smoked T41

6.3 Relative risk (95% CI) of lung cancer in relation to lowest vs highest tar level T45

7.1 Relative risk (95% CI) of lung cancer in hand rolled vs manufactured cigarette smokers (current + former smokers, all cell types) T47

7.2 Meta-analyses for hand rolled vs manufactured T50

7.3 Relative risk (95% CI) of lung cancer for hand rolled compared to manufactured cigarette smokers - by histological type T51

8.1 Relative risk (95% CI) of lung cancer for smokers of black (dark) cigarettes compared to smokers of blond (light) cigarettes (current + former smokers, all cell types) T52

8.2 Meta-analyses for black (dark) vs blond (light) T55

8.3 Relative risk (95% CI) of lung cancer for ever smokers of black (dark) cigarettes compared to smokers of blond (light) cigarettes only - by histological type T56

9.1 Relative risk (95% CI) of lung cancer in mentholated vs nonmentholated cigarette smokers T57

9.2 Mentholated cigarettes - meta-analysis of results for regular use T59

9.3 Relative risk (95% CI) of lung cancer by nicotine level of brand smoked T60

9.4 Relative risk (95% CI) of lung cancer in bidi vs cigarettes smokers T61

9.5 Relative risk (95% CI) of lung cancer in smokers of brands local and not local in Okinawa T62

9.6 Relative risk (95% CI) of lung cancer in pilli vs pölli smokers T63

10.1 Summary of meta-analyses for major cigarette type comparisons T64

Appendices

A. Summary of each study and its main results

B. References to the 54 studies

 

Back