- Email: [email protected]
Coffee consumption and risk of lung cancer: A meta-analysis
Lung Cancer 67 (2010) 17–22
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Coffee consumption and risk of lung cancer: A meta-analysis Naping Tang a,∗ , Yuemin Wu b , Jing Ma a,∗ , Bin Wang c , Rongbin Yu d a National Shanghai Center for New Drug Safety Evaluation and Research, Shanghai Institute of Pharmaceutical Industry, 199 Guoshoujing Road, Zhangjiang Hi-Tech Park, Pudong, Shanghai 201203, China b Department of General Surgery, People’s Hospital of Liyang City, Liyang, Jiangsu Province, China c Department of Pharmacology, Nanjing Medical University, Nanjing, Jiangsu Province, China d Department of Epidemiology and Biostatistics, School of Public Health, Nanjing Medical University, Nanjing, Jiangsu Province, China
a r t i c l e
i n f o
Article history: Received 7 January 2009 Received in revised form 8 March 2009 Accepted 11 March 2009 Keywords: Coffee Lung cancer Meta-analysis
a b s t r a c t Epidemiologic studies have evaluated the potential association between coffee consumption and lung cancer risk. However, results were inconsistent. To clarify the role of coffee in lung cancer, we conducted a meta-analysis on this topic. We searched PubMed and EMBASE databases (from 1966 to January 2009) and the reference lists of retrieved articles. Study-specific risk estimates were pooled using random-effects model. Five prospective studies and 8 case–control studies involving 5347 lung cancer cases and 104,911 non-cases were included in this meta-analysis. The combined results indicated a significant positive association between highest coffee intake and lung cancer [relative risk (RR) = 1.27, 95% confidence interval (CI) = 1.04–1.54). Furthermore, an increase in coffee consumption of 2 cups/day was associated with a 14% increased risk of developing lung cancer (RR = 1.14, 95% CI = 1.04–1.26). In stratified analyses, the highest coffee consumption was significantly associated with increased risk of lung cancer in prospective studies, studies conducted in America and Japan, but borderline significantly associated with decreased risk of lung cancer in non-smokers. In addition, decaffeinated coffee drinking was associated with decreased lung cancer risk, although the number of studies on this topic was relative small. In conclusion, results from this meta-analysis indicate that high or an increased consumption of coffee may increase the risk of lung cancer. Because the residual confounding effects of smoking or other factors may still exist, these results should be interpreted with caution. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Lung cancer is the most common cancer worldwide and also is the leading cause of cancer mortality for both men and women [1,2]. According to American Cancer Society statistics, it was diagnosed in 1.5 million new cases of lung cancer by 2007 worldwide, and 1.35 million people died of the disease [2]. There is currently no effective means to screen for this cancer, and the 5-year survival rates have remained under 20% [3]. Consequently, identification of modifiable risk factors for lung cancer is of importance as it may lead to prevention opportunities. Cigarette smoking is one of the few accepted modifiable risk factors for lung cancer, but global statistics estimate that 15% of lung cancers in men and 53% in women are not attributable to smoking, overall accounting for 25% of all lung cancer cases worldwide [1]. Coffee is one of the most widely consumed beverages in the world. It contains complex mixtures of biochemically active
∗ Corresponding author. Tel.: +86 21 50800333; fax: +86 21 50801259. E-mail addresses: [email protected] (N. Tang), [email protected] (J. Ma). 0169-5002/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2009.03.012
components that have been hypothesized to influence cancer risk. These constituents have been described as not only having antimutagenic and antioxidant activities and the ability to inhibit cancer-promoting agents such as kahweol and cafestol palmitates [4], but also having genotoxic and mutagenic properties such as caffeine [5]. Epidemiologic studies also reported inconsistent findings for coffee consumption and many cancer types, especially for lung cancer. Therefore, we performed a meta-analysis to summarize the available evidence from prospective and case–control studies on the association between coffee consumption and lung cancer. 2. Methods 2.1. Search strategy We searched PubMed and EMBASE (from 1966 to January 2009) for all relevant papers, using the keywords and/or Medical Subject Headings “lung neoplasm”, “lung cancer”, “lung tumor” or “lung carcinoma” combined with “coffee”, “caffeine”. Moreover, we searched any additional studies in the references lists of retrieved articles. No language restrictions were imposed.
18
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Fig. 1. Risk estimates from studies assessing the association between high coffee consumption (highest versus non/lowest) and lung cancer risk. Squares indicated studyspecific risk estimates (size of square reflects the study-statistical weight, i.e. inverse of variance); horizontal lines indicate 95% confidence intervals; diamond indicates summary relative risk estimate with its corresponding 95% confidence interval.
2.2. Study selection Studies were included in the meta-analysis if they met following criteria: First, they had a prospective or case–control design. Second, the exposure of interest was coffee consumption. Third, the outcome of interest was primary lung cancer. Fourth, the study had to enroll “unrelated subjects” (there is no genetic correlation among the subjects included in study population). Last, the relative risk (RR) estimates [odds ratios (ORs) in case–control studies] with their corresponding 95% confidence intervals (CIs) were reported (or sufficient data to calculate them). If data were duplicated in more than one study, the most recent and complete study was included in the analysis. We identified 20 potentially relevant studies [6–25] concerning coffee consumption and lung cancer. Seven studies were excluded for following reasons: insufficient data were provided in the study to calculate 95% CIs [6], did not publish data regarding coffee consumption and lung cancer independently [7], or the control subjects enrolled in the study (visitors and patients’ bystanders) were related to the cases [8]. The other 4 studies [9–12] were excluded because they were updated by Kubik et al. [25] in 2008. Thus, thirteen studies [13–25] on coffee consumption and lung cancer risk were included in our meta-analysis. 2.3. Data extraction We extracted the following information from each study: the first author’s last name, publication year, study design, study population, type of controls for case–control studies (population-based or hospital-based controls), study period, sample size, coffee type, the exposure of the coffee consumption, the RRs or ORs and their 95% CIs for every category of coffee consumption and covariates for adjustment in the analysis. If 95% CIs were not reported, but numbers of cases and controls (or person-time) for each category of coffee consumption were provided, these data were used to calculate a standard error of the crude OR/RR, and then approximate CIs for the reported adjusted ORs/RRs. For each study we extracted the
risk estimates that reflected the greatest degree of control potential confounders. 2.4. Statistical analysis The measure of interest was the RR for prospective studies, approximated by OR in case–control studies, and the corresponding 95% CI. We estimated the risk of lung cancer for highest coffee consumption compared with non/lowest coffee consumption. Study-specified RRs and corresponding 95% CIs for highest versus non/lowest coffee consumption levels were extracted, and then log RRs were weighted by the inverse of their variances to obtain a pooled RR and its 95% CI. Studies were combined using the DerSimonian and Laird random-effect model, which incorporates both within- and between-study variability [26]. We also performed meta-analysis of the dose–response association between coffee consumption and lung cancer risk. The method proposed by Greenland and Longnecker [27] and Orsini et al. [28] was used to estimate study-specific slopes from the correlated natural logarithm of the RR across categories of coffee consumption. We attempted to place the studies on a common scale by estimating the RR for an increase in coffee consumption by 2 cups/day. For each study, we estimated the median coffee consumption for each category by assigning the midpoint of the upper and lower boundary in each category as the average consumption. Because the higher category of consumption was usually open, we considered it of the same amplitude as the preceding category [27]. This method requires the risk estimates with their variance estimates for 3 or more quantitative exposure categories. Thus, studies [13,16,17,25] with only two categories were not included in this analysis. This method also requires the number of cases and controls (or person-time) for each category. When this information was not available [23], we estimated the dose–response slopes using variance-weighted least squares regression analysis. Statistical heterogeneity among studies was evaluated using the Q and I2 statistics [29]. To avoid type II errors resulting from low power, statistical significance was defined as P < 0.10 rather
Table 1 Characteristics of studies included in the meta-analysis. Study
Study design
Study population
Study period
Cases/cohort or controls
Coffee consumption
RR/OR (95% CI)
Adjustments
Jacobsen et al. (1986)
Prospective
Norway
1967–1978
177/16,555
≤2cups/day ≥7cups/day
1.00 (reference) 1.82 (1.03–2.94)
Sex, age and residence
Nomura et al. (1986)
Prospective
Japan
1965–1983
110/7355
0 cup/day
1.00 (reference)
Age, year of smoking, number of cigarettes smoked per day, smoking status at exam, and past smoking status
1–2 cups/day 3–4 cups/day ≥5 cups/day
1.05 (0.60–2.59) 1.05 (0.65–2.90) 1.44 (1.15–4.63)
Prospective
Norway
1977–1990
125/43,000
≤4 cups/day 5–6 cups/day ≥7 cups/day
1.00 (reference) 1.54 (0.87–2.72) 2.29 (1.38–3.80)
Age, sex, cigarettes per day and county of residence
Fu et al. (1997)
Prospective
Japan
1985–1995
161/24,489
Non-drinkers
1.00 (reference)
Smoking status, tea intake, yellow or green vegetables consumption, fruit consumption, and wine drinking
Ever drinkers
1.68 (1.17–2.40)
Khan et al. (2004)
Prospective
Japan
1984–2002
51/3158
several times/month
1.00 (reference) 0.84 (0.48–1.49)
Age and smoking
Axelsson et al. (1996)
PCC
Sweden
1989–1993
308/504
<2 times/week
1.00 (reference)
No. of cigarettes/day, no. of years smoked, marital status, socioeconomic job classification, vegetable class and other fruit or berries
Daily/almost daily 7–25 times/week >25 times/week
0.94 (0.38–2.29) 1.16 (0.53–2.52) 1.60 (0.72–3.54)
Less than daily
1.00 (reference)
Daily/almost daily ≥3 cups/day
0.69 (0.33–1.43) 0.55 (0.27–1.12)
≤1 cup/week 2–7 cups/week 8–17.5 cups/week >17.5 cups/week Non-drinkers <1 cup/day 2–3 cups/day ≥4 cups/day
1.00 (reference) 0.90 (0.50–1.60) 0.90 (0.50–1.60) 0.80 (0.40–1.80) 1.00 (reference) 1.01 (0.67–1.51) 0.94 (0.65–1.37) 1.26 (0.86–1.84)
Non-drinkers
1.00 (reference)
1 cup/week 2–3 cups/week 1 cup/day ≥2 cups/day
1.32 (0.75–2.33) 0.88 (0.49–1.55) 1.20 (0.60–2.41) 1.22 (0.53–2.80)
<1 cup/day
1.00 (reference)
1 cup/day 2 cups/day ≥ 3 cups/day
0.87 (0.72–1.07) 0.94 (0.75–1.18) 1.32 (1.02–1.70)
Non-drinkers
1.00 (reference)
≤1 cup/day 2–3 cups/day ≥4 cups/day
1.03 (0.73–1.45) 1.34 (0.99–1.82) 1.51 (1.11–2.05)
Non-drinkers
1.00 (reference)
Nyberg et al. (1998)
PCC
Sweden
1989–1995
124/2235
Hu et al. (2002)
PCC
Canada
1994–1997
161/483
Mettlin et al. (1989)
HCC
United States
1982–1987
569/569
Mendilaharsu et al. (1998)
HCC
Uruguay
1994–1996
427/428
Takezaki et al. (2001)
Baker et al. (2005)
Kubik et al. (2008)
HCC
HCC
HCC
Japan
United States
Czech
1988–1997
1982–1998
1998–2006
1045/4153
993/986
1096/2996
Age, province, education and social class
Sex, smoking history, beta-carotene intake index and education level
Age, residence, urban/rural status, tobacco smoking, total energy intake, dairy foods, desserts, all vegetables and fruits, mate intake, caffeine index, and coffee
Age, season and year of visit, occupation, prior lung disease, smoking and consumption of green vegetables and meat
Age, sex, smoking status, known occupational exposure to other kinds of dust, known occupational exposure to smoke, number of cigarettes smoked per day, and interaction between smoke exposure and cigarettes
Age, residence, education and pack-years of smoking
19
CI, confidence interval; HCC, hospital-based case–control study; OR, odds ratio; PCC, population-based case–control study; RR, relative risk.
Age, gender, catchment, smoking, degree of urban residence, years of exposure to risk occupations, ever-exposure status, years since last exposure and hour-year of exposure to environmental tobacco smoke, carrot consumption and other fruits consumption
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Stensvold et al. (1994)
20
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Table 2 Summary risk estimates for coffee consumption (highest versus non/lowest) and lung cancer risk. Study
All studies Study design Prospective studies Case–control studies Population-based controls Hospital-based controls
No. of studies
RR (95% CI)
Heterogeneity test Q
P
I2 (%)
13
1.27 (1.04–1.54)
28.99
0.004
58.6
5 8 3 5
1.57 (1.15–2.14) 1.13 (0.90–1.41) 0.87 (0.48–1.60) 1.20 (0.95–1.51)
7.29 15.04 3.89 9.28
0.121 0.036 0.143 0.054
45.1 53.5 48.5 56.9
Study population Americaa Japan Europeb
4 4 5
1.33 (1.07–1.65) 1.34 (1.05–1.70) 1.26 (0.76–2.09)
2.53 4.18 19.61
0.470 0.242 0.001
0.0 28.3 79.6
Smoking status Non-smoking smoking
3 2
0.78 (0.60–1.00) 1.28 (0.87–1.88)
1.04 2.13
0.594 0.144
0.0 53.1
Histological subtype Adenocarcinomas Squamous cell & small cell carcinomas
3 4
1.18 (0.94–1.49) 1.23 (0.89–1.71)
1.70 7.82
0.428 0.050
0.0 61.6
Coffee type Decaffeinated coffee
2
0.66 (0.54–0.81)
0.54
0.461
0.0
CI, confidence interval; RR, relative risk. a Including 2 studies conducted in United States, 1 in Canada, and 1 in Uruguay. b Including 2 studies conducted in Norway, 2 in Sweden and 1 in Czech.
than the traditional 0.05 [30]. When statistical heterogeneity was detected, the sources of heterogeneity were explored and sensitivity analysis was performed. Publication bias was evaluated through visual inspection of funnel plots, and the Egger weighted regression method with P < 0.10 was considered representative of statistical significance [31]. All statistical analyses were performed with Stata (version 9.0; StataCorp, College Station, TX). 3. Results We identified 5 prospective [13–17] and 8 case–control studies [18–25] of association between coffee consumption and risk of lung cancer (Table 1). Of these studies, 2 were conducted in Norway [13,15], 4 in Japan [14,16,17,23], 2 in Sweden [18,19], 2 in United States [21,24], 1 in Canada [20], 1 in Uruguay [22] and 1 in Czech [25]. Of the 8 case–control studies, 3 used population-based controls [18–20] and 5 used hospital-based controls [21–25]. The estimated RRs of lung cancer for highest coffee consumption compared with non/lowest coffee consumption are presented in Fig. 1 and Table 2. Overall, the summary RR indicated that highest coffee consumption was statistically significantly associated with an increased risk of lung cancer (RR = 1.27, 95% CI = 1.04–1.54). There was significant heterogeneity across the studies (Q = 28.99, P = 0.004, I2 = 58.6%). By using a stepwise process, we noted that most of the heterogeneity was accounted for 2 studies by Nyberg et al. [19] and Kubik et al. [25]. When these 2 studies were excluded, the summary estimate was essentially unchanged (RR = 1.42, 95% CI = 1.23–1.65), but a concomitant shift in heterogeneity was measured by Q test (from P = 0.004 to P = 0.304). No evidence of publication bias was observed from either visualization of the funnel plot (Fig. 2) or Egger’s test (P = 0.878). When subgroup analysis was conducted by study design, a statistically significant 57% increased risk of developing lung cancer was observed among prospective studies (RR = 1.57, 95% CI = 1.15–2.14; Q = 7.29, P = 0.121, I2 = 45.1%), while no significant association was observed among case–control studies (RR = 1.13, 95% CI = 0.90–1.41; Q = 15.04, P = 0.036, I2 = 53.5%) (Table 2). When stratified the various studies by study population, statistically significant association between high coffee consumption and lung cancer risk was observed among studies conducted in Amer-
ica (RR = 1.33, 95% CI = 1.07–1.65; Q = 2.53, P = 0.470, I2 = 0.0%) and Japan (RR = 1.34, 95% CI = 1.05–1.70; Q = 4.18, P = 0.242, I2 = 28.3%). However, no significant association was noted among studies conducted in Europe (RR = 1.26, 95% CI = 0.76–2.09; Q = 19.61, P = 0.001, I2 = 79.6%) (Table 2). We also examined the coffee–lung cancer association in smokers and non-smokers separately. The overall results in smokers indicated a non-significant positive association between high coffee consumption and lung cancer risk (RR = 1.28, 95% CI = 0.87–1.88;
Fig. 2. Funnel plot indicating publication bias in the studies included in this metaanalysis. No indication of publication bias was noted from both visualization of funnel plot and the Egger’s test (P = 0.878).
N. Tang et al. / Lung Cancer 67 (2010) 17–22
21
Fig. 3. Risk estimates from studies assessing the association between an increment of coffee consumption of 2 cups/day and lung cancer risk. Squares indicated study-specific risk estimates (size of square reflects the study-statistical weight, i.e. inverse of variance); horizontal lines indicate 95% confidence intervals; diamond indicates summary relative risk estimate with its corresponding 95% confidence interval.
Q = 2.13, P = 0.144, I2 = 53.1%). However, a borderline significant inverse association was noted in non-smokers (RR = 0.78, 95% CI = 0.60–1.00; Q = 1.04, P = 0.594, I2 = 0.0%) (Table 2). Two studies have examined the association between decaffeinated coffee consumption and risk of lung cancer. Interestingly, the pooled results of these two studies found that there was a statistically significant reduction in lung cancer risk for high decaffeinated coffee consumption (RR = 0.66, 95% CI = 0.54–0.81; Q = 0.54, P = 0.461, I2 = 0.0%) (Table 2). We also examined if histological subtypes of lung cancer affected the pooled RR. However, no statistically significant association was observed in this subgroup analyses. Given the wide array of measurement categories reported in the literature, we performed a dose–response analysis and calculated a risk of lung cancer for an increase in coffee consumption of 2 cups/day (Fig. 3). Overall, we found that an increment of coffee consumption of 2 cups/day was statistically significantly associated with an 11% increased risk of developing lung cancer (RR = 1.14, 95% CI = 1.04–1.26; Q = 13.61, P = 0.093, I2 = 41.2%). 4. Discussion This meta-analysis evaluated the potential association between coffee consumption and lung cancer risk based on 5 prospective and 8 case–control studies. Overall, the summary RR indicated a significant positive association between highest coffee consumption and lung cancer risk. Furthermore, an increase in coffee consumption of 2 cups/day was statistically significantly associated with a 14% increased risk of developing lung cancer. The biologic mechanism whereby coffee increases the risk of lung cancer is likely to be multifactorial. Coffee (60–75%) is the major source of caffeine in diet. Previous studies have demonstrated that caffeine can affect DNA repair, modify the apoptotic response and perturb cell cycle checkpoint integrity [32–38]. It can also influence the normal induction of p53 in response to DNA damage through modifying p53 status [37,38]. In addition, study by Liu [39] has shown that caffeine can reduce the antioxidant and anticancer activities of flavonoids which have been suggested to play an
important role against the risk of lung cancer [40]. Furthermore, caffeine can induce the production of cytochrome P450 (CYP) enzymes such as CYP1A1/1A2 [41]. CYP1A1 was reported to activate a number of chemical carcinogens, including polycyclic hydrocarbons and their oxygenated derivatives, heterocyclic amines, aromatic amines, and nitropolycylic hydrocarbons [42]. Although the effect of coffee on lung cancer is biologically plausible, we could not rule out the possibility that the association might be confounded by smoking status. Cigarette smoking is one of the established risk factors for lung cancer. It has been reported that high intakes of coffee are frequently associated with cigarette smoking [43]. In addition, the stratified analysis by smoking status in this study indicated a borderline significant inverse association between coffee consumption and lung cancer among non-smokers but a non-significant positive association among smokers, though the non-significant positive association may be by chance because only 2 studies were included in this analysis. Therefore, our findings of a positive association between coffee consumption and lung cancer risk should be interpreted with caution. When subgroup analysis was conducted by study design, we noted that statistically significant positive association was only observed among prospective studies but not in case–control studies. This may be due to the differential misclassification of coffee consumption in case–control studies, because the information on coffee consumption was collected after the patients were diagnosed. Thus, the collected consumption data may not reflect relevant exposures considering the long latency of lung cancer. In stratified analysis by study population, we found that coffee consumption was associated with increased risk of lung cancer in studies conducted in America and Japan but not in studies conducted in Europe. The different observations may be explained at least in part, by the variations of coffee consumption or preparation methods among different countries. In addition, a possible role of ethnic differences in genetic backgrounds and the environment they lived in should also be taken into consideration. Interestingly, when we pooled the 2 studies on the association of decaffeinated coffee with lung cancer, we observed a protective effect of high decaffeinated coffee consumption on lung cancer.
22
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Decaffeinated coffee contains very little caffeine, but does not reduce the levels of cafestol and kahweol. Importantly, cafestol and kahweol have been shown to produce a broad range of biochemical effects resulting in a reduction of genotoxicity of carcinogens [4]. In addition, it is also likely that the observed effect of decaffeinated coffee on lung cancer may be due to chance because only 2 studies were included in the analysis. Therefore, our results should be interpreted with caution. Some limitations of this meta-analysis should be acknowledged. First, only observational studies were included in our meta-analysis. Observational studies, even when well controlled, are susceptible to various biases. However, prospective studies, which are less susceptible to bias because of the prospective design, also showed a positive association between coffee consumption and lung cancer, suggesting the finding is not likely attributable to recall and selection bias. Second, our results are likely to be affected by some misclassification of coffee consumption. Coffee exposure is mostly assessed regarding the number of cups of coffee consumed daily, weekly or monthly. However, cup size may vary considerably. Third, we extracted the risk estimates that reflected the greatest degree of the control potential confounders, because it was hard to obtain raw data from each study for conducting standardized adjustments. Therefore, it was probably that the results based on the adjustment for different confounders were different from those based on standardized adjustments. Finally, only published studies were included in our meta-analysis. Therefore, publication bias may have occurred although no publication bias was indicated from both visualization of the funnel plot and Egger’s test. 5. Conclusions In summary, this meta-analysis of 5 prospective and 8 case–control studies involving 5347 lung cancer cases and 104,911 non-cases suggests that increased consumption of coffee may increase the risk of developing lung cancer. To provide a more definitive conclusion, further pooled analyses with more complete raw data or prospective cohort studies with larger sample size, well controlled confounding factors and longer duration of follow-up are needed in this area. Conflict of interest statement The authors promised there were not any possible conflicts of interest in this research. References [1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108. [2] Garcia M, Jemal A, Ward EM, Center MM, Hao Y, Siegel RL, et al. Global cancer facts & figures 2007. Atlanta, GA: American Cancer Society; 2007. [3] Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, et al. Cancer statistics, 2008. CA Cancer J Clin 2008;58:71–96. [4] Cavin C, Holzhaeuser D, Scharf G, Constable A, Huber WW, Schilter B. Cafestol and kahweol, two coffee specific diterpenes with anticarcinogenic activity. Food Chem Toxicol 2002;40:1155–63. [5] Deplanque G, Ceraline J, Mah-Becherel MC, Cazenave JP, Bergerat JP, Klein-Soyer C. Caffeine and the G2/M block override: a concept resulting from a misleading cell kinetic delay, independent of functional p53. Int J Cancer 2001;94:363–9. [6] Restrepo HE, Correa P, Haenszel W, Brinton LA, Franco A. A case–control study of tobacco-related cancers in Colombia. Bull Pan Am Health Organ 1989;23:405–13. [7] Zheng W, Doyle TJ, Kushi LH, Sellers TA, Hong CP, Folsom AR. Tea consumption and cancer incidence in a prospective cohort study of postmenopausal women. Am J Epidemiol 1996;144:175–82. [8] Sankaranarayanan R, Varghese C, Duffy SW, Padmakumary G, Day NE, Nair MK. A case–control study of diet and lung cancer in Kerala, south India. Int J Cancer 1994;58:644–9. [9] Kubik A, Zatloukal P, Tomasek L, Kriz J, Petruzelka L, Plesko I. Diet and the risk of lung cancer among women. A hospital-based case–control study. Neoplasma 2001;48:262–6.
[10] Kubik AK, Zatloukal P, Tomasek L, Pauk N, Havel L, Krepela E, et al. Dietary habits and lung cancer risk among non-smoking women. Eur J Cancer Prev 2004;13:471–80. [11] Kubik A, Zatloukal P, Tomasek L, Pauk N, Petruzelka L, Plesko I. Lung cancer risk among nonsmoking women in relation to diet and physical activity. Neoplasma 2004;51:136–43. [12] Kubik A, Zatloukal P, Tomasek L, Pauk N, Havel L, Dolezal J, et al. Interactions between smoking and other exposures associated with lung cancer risk in women: diet and physical activity. Neoplasma 2007;54:83–8. [13] Jacobsen BK, Bjelke E, Kvåle G, Heuch I. Coffee drinking, mortality, and cancer incidence: results from a Norwegian prospective study. J Natl Cancer Inst 1986;76:823–31. [14] Nomura A, Heilbrun LK, Stemmermann GN. Prospective study of coffee consumption and the risk of cancer. J Natl Cancer Inst 1986;76:587–90. [15] Stensvold I, Jacobsen BK. Coffee and cancer: a prospective study of 43,000 Norwegian men and women. Cancer Causes Control 1994;5:401–8. [16] Fu YY, Takezaki T, Tajima K. Risk factors of lung cancer—follow-up studies in Nagoya Japan. Zhonghua Liu Xing Bing Xue Za Zhi 1997;18:328–30. [17] Khan MM, Goto R, Kobayashi K, Suzumura S, Nagata Y, Sonoda T, et al. Dietary habits and cancer mortality among middle aged and older Japanese living in Hokkaido, Japan by cancer site and sex. Asian Pac J Cancer Prev 2004;5:58–65. [18] Axelsson G, Liljeqvist T, Andersson L, Bergman B, Rylander R. Dietary factors and lung cancer among men in west Sweden. Int J Epidemiol 1996;25:32–9. [19] Nyberg F, Agrenius V, Svartengren K, Svensson C, Pershagen G. Dietary factors and risk of lung cancer in never-smokers. Int J Cancer 1998;78:430–6. [20] Hu J, Mao Y, Dryer D, White K. Risk factors for lung cancer among Canadian women who have never smoked. Cancer Detect Prev 2002;26:129–38. [21] Mettlin C. Milk drinking, other beverage habits, and lung cancer risk. Int J Cancer 1989;43:608–12. [22] Mendilaharsu M, De Stefani E, Deneo-Pellegrini H, Carzoglio JC, Ronco A. Consumption of tea and coffee and the risk of lung cancer in cigarette-smoking men: a case–control study in Uruguay. Lung Cancer 1998;19:101–7. [23] Takezaki T, Hirose K, Inoue M, Hamajima N, Yatabe Y, Mitsudomi T, et al. Dietary factors and lung cancer risk in Japanese: with special reference to fish consumption and adenocarcinomas. Br J Cancer 2001;84:1199–206. [24] Baker JA, McCann SE, Reid ME, Nowell S, Beehler GP, Moysich KB. Associations between black tea and coffee consumption and risk of lung cancer among current and former smokers. Nutr Cancer 2005;52:15–21. [25] Kubik A, Zatloukal P, Tomasek L, Dolezal J, Syllabova L, Kara J, et al. A case–control study of lifestyle and lung cancer associations by histological types. Neoplasma 2008;55:192–9. [26] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–88. [27] Greenland S, Longnecker MP. Methods for trend estimation from summarized dose–response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301–9. [28] Orsini N, Bellocco R, Greenland S. Generalized least squares for trend estimation of summarized dose–response data. Stata J 2006;6:40–57. [29] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58. [30] Hedges LV, Pigott TD. The power of statistical tests in metaanalysis. Psychol Methods 2001;6:203–17. [31] Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–34. [32] Mohr U, Emura M, Riebe-Imre M. Experimental studies on carcinogenicity and mutagenicity of caffeine. In: Garattini S, editor. Caffeine, coffee, and health. New York: Raven Press; 1993. p. 359–78. [33] Agapova LS, Ilyinskaya GV, Turovets NA, Ivanov AV, Chumakov PM, Kopnin BP. Chromosome changes caused by alterations of p53 expression. Mutat Res 1996;354:129–38. [34] DeFrank JS, Tang W, Powell SN. p53-null cells are more sensitive to ultraviolet light only in the presence of caffeine. Cancer Res 1996;56:5365–8. [35] Efferth T, Fabry U, Glatte P, Osieka R. Expression of apoptosis-related oncoproteins and modulation of apoptosis by caffeine in human leukemic cells. J Cancer Res Clin Oncol 1995;121:648–56. [36] Link Jr CJ, Evans MK, Cook JA, Muldoon R, Stevnsner T, Bohr VA. Caffeine inhibits gene-specific repair of UV-induced DNA damage in hamster cells and in human xeroderma pigmentosum group C cells. Carcinogenesis 1995;16:1149–55. [37] Müller WU, Bauch T, Wojcik A, Böcker W, Streffer C. Comet assay studies indicate that caffeine-mediated increase in radiation risk of embryos is due to inhibition of DNA repair. Mutagenesis 1996;11:57–60. [38] Ito K, Nakazato T, Miyakawa Y, Yamato K, Ikeda Y, Kizaki M. Caffeine induces G2/M arrest and apoptosis via a novel p53-dependent pathway in NB4 promyelocytic leukemia cells. J Cell Physiol 2003;196:276–83. [39] Liu RH. Health benefits of fruit and vegetables are from additive and synergistic combinations of phytochemicals. Am J Clin Nutr 2003;78:517S–20S. [40] Arts IC. A review of the epidemiological evidence on tea, flavonoids, and lung cancer. J Nutr 2008;138:1561S–6S. [41] Goasduff T, Dréano Y, Guillois B, Ménez JF, Berthou F. Induction of liver and kidney CYP1A1/1A2 by caffeine in rat. Biochem Pharmacol 1996;52:1915–9. [42] Nebert DW, Dalton TP, Okey AB, Gonzalez FJ. Role of aryl hydrocarbon receptormediated induction of the CYP1 enzymes in environmental toxicity and cancer. J Biol Chem 2004;279:23847–50. [43] Leviton A, Cowan L. A review of the literature relating caffeine consumption by women to their risk of reproductive hazards. Food Chem Toxicol 2002;40:1271–310.
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Coffee consumption and risk of lung cancer: A meta-analysis Naping Tang a,∗ , Yuemin Wu b , Jing Ma a,∗ , Bin Wang c , Rongbin Yu d a National Shanghai Center for New Drug Safety Evaluation and Research, Shanghai Institute of Pharmaceutical Industry, 199 Guoshoujing Road, Zhangjiang Hi-Tech Park, Pudong, Shanghai 201203, China b Department of General Surgery, People’s Hospital of Liyang City, Liyang, Jiangsu Province, China c Department of Pharmacology, Nanjing Medical University, Nanjing, Jiangsu Province, China d Department of Epidemiology and Biostatistics, School of Public Health, Nanjing Medical University, Nanjing, Jiangsu Province, China
a r t i c l e
i n f o
Article history: Received 7 January 2009 Received in revised form 8 March 2009 Accepted 11 March 2009 Keywords: Coffee Lung cancer Meta-analysis
a b s t r a c t Epidemiologic studies have evaluated the potential association between coffee consumption and lung cancer risk. However, results were inconsistent. To clarify the role of coffee in lung cancer, we conducted a meta-analysis on this topic. We searched PubMed and EMBASE databases (from 1966 to January 2009) and the reference lists of retrieved articles. Study-specific risk estimates were pooled using random-effects model. Five prospective studies and 8 case–control studies involving 5347 lung cancer cases and 104,911 non-cases were included in this meta-analysis. The combined results indicated a significant positive association between highest coffee intake and lung cancer [relative risk (RR) = 1.27, 95% confidence interval (CI) = 1.04–1.54). Furthermore, an increase in coffee consumption of 2 cups/day was associated with a 14% increased risk of developing lung cancer (RR = 1.14, 95% CI = 1.04–1.26). In stratified analyses, the highest coffee consumption was significantly associated with increased risk of lung cancer in prospective studies, studies conducted in America and Japan, but borderline significantly associated with decreased risk of lung cancer in non-smokers. In addition, decaffeinated coffee drinking was associated with decreased lung cancer risk, although the number of studies on this topic was relative small. In conclusion, results from this meta-analysis indicate that high or an increased consumption of coffee may increase the risk of lung cancer. Because the residual confounding effects of smoking or other factors may still exist, these results should be interpreted with caution. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Lung cancer is the most common cancer worldwide and also is the leading cause of cancer mortality for both men and women [1,2]. According to American Cancer Society statistics, it was diagnosed in 1.5 million new cases of lung cancer by 2007 worldwide, and 1.35 million people died of the disease [2]. There is currently no effective means to screen for this cancer, and the 5-year survival rates have remained under 20% [3]. Consequently, identification of modifiable risk factors for lung cancer is of importance as it may lead to prevention opportunities. Cigarette smoking is one of the few accepted modifiable risk factors for lung cancer, but global statistics estimate that 15% of lung cancers in men and 53% in women are not attributable to smoking, overall accounting for 25% of all lung cancer cases worldwide [1]. Coffee is one of the most widely consumed beverages in the world. It contains complex mixtures of biochemically active
∗ Corresponding author. Tel.: +86 21 50800333; fax: +86 21 50801259. E-mail addresses: [email protected] (N. Tang), [email protected] (J. Ma). 0169-5002/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2009.03.012
components that have been hypothesized to influence cancer risk. These constituents have been described as not only having antimutagenic and antioxidant activities and the ability to inhibit cancer-promoting agents such as kahweol and cafestol palmitates [4], but also having genotoxic and mutagenic properties such as caffeine [5]. Epidemiologic studies also reported inconsistent findings for coffee consumption and many cancer types, especially for lung cancer. Therefore, we performed a meta-analysis to summarize the available evidence from prospective and case–control studies on the association between coffee consumption and lung cancer. 2. Methods 2.1. Search strategy We searched PubMed and EMBASE (from 1966 to January 2009) for all relevant papers, using the keywords and/or Medical Subject Headings “lung neoplasm”, “lung cancer”, “lung tumor” or “lung carcinoma” combined with “coffee”, “caffeine”. Moreover, we searched any additional studies in the references lists of retrieved articles. No language restrictions were imposed.
18
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Fig. 1. Risk estimates from studies assessing the association between high coffee consumption (highest versus non/lowest) and lung cancer risk. Squares indicated studyspecific risk estimates (size of square reflects the study-statistical weight, i.e. inverse of variance); horizontal lines indicate 95% confidence intervals; diamond indicates summary relative risk estimate with its corresponding 95% confidence interval.
2.2. Study selection Studies were included in the meta-analysis if they met following criteria: First, they had a prospective or case–control design. Second, the exposure of interest was coffee consumption. Third, the outcome of interest was primary lung cancer. Fourth, the study had to enroll “unrelated subjects” (there is no genetic correlation among the subjects included in study population). Last, the relative risk (RR) estimates [odds ratios (ORs) in case–control studies] with their corresponding 95% confidence intervals (CIs) were reported (or sufficient data to calculate them). If data were duplicated in more than one study, the most recent and complete study was included in the analysis. We identified 20 potentially relevant studies [6–25] concerning coffee consumption and lung cancer. Seven studies were excluded for following reasons: insufficient data were provided in the study to calculate 95% CIs [6], did not publish data regarding coffee consumption and lung cancer independently [7], or the control subjects enrolled in the study (visitors and patients’ bystanders) were related to the cases [8]. The other 4 studies [9–12] were excluded because they were updated by Kubik et al. [25] in 2008. Thus, thirteen studies [13–25] on coffee consumption and lung cancer risk were included in our meta-analysis. 2.3. Data extraction We extracted the following information from each study: the first author’s last name, publication year, study design, study population, type of controls for case–control studies (population-based or hospital-based controls), study period, sample size, coffee type, the exposure of the coffee consumption, the RRs or ORs and their 95% CIs for every category of coffee consumption and covariates for adjustment in the analysis. If 95% CIs were not reported, but numbers of cases and controls (or person-time) for each category of coffee consumption were provided, these data were used to calculate a standard error of the crude OR/RR, and then approximate CIs for the reported adjusted ORs/RRs. For each study we extracted the
risk estimates that reflected the greatest degree of control potential confounders. 2.4. Statistical analysis The measure of interest was the RR for prospective studies, approximated by OR in case–control studies, and the corresponding 95% CI. We estimated the risk of lung cancer for highest coffee consumption compared with non/lowest coffee consumption. Study-specified RRs and corresponding 95% CIs for highest versus non/lowest coffee consumption levels were extracted, and then log RRs were weighted by the inverse of their variances to obtain a pooled RR and its 95% CI. Studies were combined using the DerSimonian and Laird random-effect model, which incorporates both within- and between-study variability [26]. We also performed meta-analysis of the dose–response association between coffee consumption and lung cancer risk. The method proposed by Greenland and Longnecker [27] and Orsini et al. [28] was used to estimate study-specific slopes from the correlated natural logarithm of the RR across categories of coffee consumption. We attempted to place the studies on a common scale by estimating the RR for an increase in coffee consumption by 2 cups/day. For each study, we estimated the median coffee consumption for each category by assigning the midpoint of the upper and lower boundary in each category as the average consumption. Because the higher category of consumption was usually open, we considered it of the same amplitude as the preceding category [27]. This method requires the risk estimates with their variance estimates for 3 or more quantitative exposure categories. Thus, studies [13,16,17,25] with only two categories were not included in this analysis. This method also requires the number of cases and controls (or person-time) for each category. When this information was not available [23], we estimated the dose–response slopes using variance-weighted least squares regression analysis. Statistical heterogeneity among studies was evaluated using the Q and I2 statistics [29]. To avoid type II errors resulting from low power, statistical significance was defined as P < 0.10 rather
Table 1 Characteristics of studies included in the meta-analysis. Study
Study design
Study population
Study period
Cases/cohort or controls
Coffee consumption
RR/OR (95% CI)
Adjustments
Jacobsen et al. (1986)
Prospective
Norway
1967–1978
177/16,555
≤2cups/day ≥7cups/day
1.00 (reference) 1.82 (1.03–2.94)
Sex, age and residence
Nomura et al. (1986)
Prospective
Japan
1965–1983
110/7355
0 cup/day
1.00 (reference)
Age, year of smoking, number of cigarettes smoked per day, smoking status at exam, and past smoking status
1–2 cups/day 3–4 cups/day ≥5 cups/day
1.05 (0.60–2.59) 1.05 (0.65–2.90) 1.44 (1.15–4.63)
Prospective
Norway
1977–1990
125/43,000
≤4 cups/day 5–6 cups/day ≥7 cups/day
1.00 (reference) 1.54 (0.87–2.72) 2.29 (1.38–3.80)
Age, sex, cigarettes per day and county of residence
Fu et al. (1997)
Prospective
Japan
1985–1995
161/24,489
Non-drinkers
1.00 (reference)
Smoking status, tea intake, yellow or green vegetables consumption, fruit consumption, and wine drinking
Ever drinkers
1.68 (1.17–2.40)
Khan et al. (2004)
Prospective
Japan
1984–2002
51/3158
1.00 (reference) 0.84 (0.48–1.49)
Age and smoking
Axelsson et al. (1996)
PCC
Sweden
1989–1993
308/504
<2 times/week
1.00 (reference)
No. of cigarettes/day, no. of years smoked, marital status, socioeconomic job classification, vegetable class and other fruit or berries
Daily/almost daily 7–25 times/week >25 times/week
0.94 (0.38–2.29) 1.16 (0.53–2.52) 1.60 (0.72–3.54)
Less than daily
1.00 (reference)
Daily/almost daily ≥3 cups/day
0.69 (0.33–1.43) 0.55 (0.27–1.12)
≤1 cup/week 2–7 cups/week 8–17.5 cups/week >17.5 cups/week Non-drinkers <1 cup/day 2–3 cups/day ≥4 cups/day
1.00 (reference) 0.90 (0.50–1.60) 0.90 (0.50–1.60) 0.80 (0.40–1.80) 1.00 (reference) 1.01 (0.67–1.51) 0.94 (0.65–1.37) 1.26 (0.86–1.84)
Non-drinkers
1.00 (reference)
1 cup/week 2–3 cups/week 1 cup/day ≥2 cups/day
1.32 (0.75–2.33) 0.88 (0.49–1.55) 1.20 (0.60–2.41) 1.22 (0.53–2.80)
<1 cup/day
1.00 (reference)
1 cup/day 2 cups/day ≥ 3 cups/day
0.87 (0.72–1.07) 0.94 (0.75–1.18) 1.32 (1.02–1.70)
Non-drinkers
1.00 (reference)
≤1 cup/day 2–3 cups/day ≥4 cups/day
1.03 (0.73–1.45) 1.34 (0.99–1.82) 1.51 (1.11–2.05)
Non-drinkers
1.00 (reference)
Nyberg et al. (1998)
PCC
Sweden
1989–1995
124/2235
Hu et al. (2002)
PCC
Canada
1994–1997
161/483
Mettlin et al. (1989)
HCC
United States
1982–1987
569/569
Mendilaharsu et al. (1998)
HCC
Uruguay
1994–1996
427/428
Takezaki et al. (2001)
Baker et al. (2005)
Kubik et al. (2008)
HCC
HCC
HCC
Japan
United States
Czech
1988–1997
1982–1998
1998–2006
1045/4153
993/986
1096/2996
Age, province, education and social class
Sex, smoking history, beta-carotene intake index and education level
Age, residence, urban/rural status, tobacco smoking, total energy intake, dairy foods, desserts, all vegetables and fruits, mate intake, caffeine index, and coffee
Age, season and year of visit, occupation, prior lung disease, smoking and consumption of green vegetables and meat
Age, sex, smoking status, known occupational exposure to other kinds of dust, known occupational exposure to smoke, number of cigarettes smoked per day, and interaction between smoke exposure and cigarettes
Age, residence, education and pack-years of smoking
19
CI, confidence interval; HCC, hospital-based case–control study; OR, odds ratio; PCC, population-based case–control study; RR, relative risk.
Age, gender, catchment, smoking, degree of urban residence, years of exposure to risk occupations, ever-exposure status, years since last exposure and hour-year of exposure to environmental tobacco smoke, carrot consumption and other fruits consumption
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Stensvold et al. (1994)
20
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Table 2 Summary risk estimates for coffee consumption (highest versus non/lowest) and lung cancer risk. Study
All studies Study design Prospective studies Case–control studies Population-based controls Hospital-based controls
No. of studies
RR (95% CI)
Heterogeneity test Q
P
I2 (%)
13
1.27 (1.04–1.54)
28.99
0.004
58.6
5 8 3 5
1.57 (1.15–2.14) 1.13 (0.90–1.41) 0.87 (0.48–1.60) 1.20 (0.95–1.51)
7.29 15.04 3.89 9.28
0.121 0.036 0.143 0.054
45.1 53.5 48.5 56.9
Study population Americaa Japan Europeb
4 4 5
1.33 (1.07–1.65) 1.34 (1.05–1.70) 1.26 (0.76–2.09)
2.53 4.18 19.61
0.470 0.242 0.001
0.0 28.3 79.6
Smoking status Non-smoking smoking
3 2
0.78 (0.60–1.00) 1.28 (0.87–1.88)
1.04 2.13
0.594 0.144
0.0 53.1
Histological subtype Adenocarcinomas Squamous cell & small cell carcinomas
3 4
1.18 (0.94–1.49) 1.23 (0.89–1.71)
1.70 7.82
0.428 0.050
0.0 61.6
Coffee type Decaffeinated coffee
2
0.66 (0.54–0.81)
0.54
0.461
0.0
CI, confidence interval; RR, relative risk. a Including 2 studies conducted in United States, 1 in Canada, and 1 in Uruguay. b Including 2 studies conducted in Norway, 2 in Sweden and 1 in Czech.
than the traditional 0.05 [30]. When statistical heterogeneity was detected, the sources of heterogeneity were explored and sensitivity analysis was performed. Publication bias was evaluated through visual inspection of funnel plots, and the Egger weighted regression method with P < 0.10 was considered representative of statistical significance [31]. All statistical analyses were performed with Stata (version 9.0; StataCorp, College Station, TX). 3. Results We identified 5 prospective [13–17] and 8 case–control studies [18–25] of association between coffee consumption and risk of lung cancer (Table 1). Of these studies, 2 were conducted in Norway [13,15], 4 in Japan [14,16,17,23], 2 in Sweden [18,19], 2 in United States [21,24], 1 in Canada [20], 1 in Uruguay [22] and 1 in Czech [25]. Of the 8 case–control studies, 3 used population-based controls [18–20] and 5 used hospital-based controls [21–25]. The estimated RRs of lung cancer for highest coffee consumption compared with non/lowest coffee consumption are presented in Fig. 1 and Table 2. Overall, the summary RR indicated that highest coffee consumption was statistically significantly associated with an increased risk of lung cancer (RR = 1.27, 95% CI = 1.04–1.54). There was significant heterogeneity across the studies (Q = 28.99, P = 0.004, I2 = 58.6%). By using a stepwise process, we noted that most of the heterogeneity was accounted for 2 studies by Nyberg et al. [19] and Kubik et al. [25]. When these 2 studies were excluded, the summary estimate was essentially unchanged (RR = 1.42, 95% CI = 1.23–1.65), but a concomitant shift in heterogeneity was measured by Q test (from P = 0.004 to P = 0.304). No evidence of publication bias was observed from either visualization of the funnel plot (Fig. 2) or Egger’s test (P = 0.878). When subgroup analysis was conducted by study design, a statistically significant 57% increased risk of developing lung cancer was observed among prospective studies (RR = 1.57, 95% CI = 1.15–2.14; Q = 7.29, P = 0.121, I2 = 45.1%), while no significant association was observed among case–control studies (RR = 1.13, 95% CI = 0.90–1.41; Q = 15.04, P = 0.036, I2 = 53.5%) (Table 2). When stratified the various studies by study population, statistically significant association between high coffee consumption and lung cancer risk was observed among studies conducted in Amer-
ica (RR = 1.33, 95% CI = 1.07–1.65; Q = 2.53, P = 0.470, I2 = 0.0%) and Japan (RR = 1.34, 95% CI = 1.05–1.70; Q = 4.18, P = 0.242, I2 = 28.3%). However, no significant association was noted among studies conducted in Europe (RR = 1.26, 95% CI = 0.76–2.09; Q = 19.61, P = 0.001, I2 = 79.6%) (Table 2). We also examined the coffee–lung cancer association in smokers and non-smokers separately. The overall results in smokers indicated a non-significant positive association between high coffee consumption and lung cancer risk (RR = 1.28, 95% CI = 0.87–1.88;
Fig. 2. Funnel plot indicating publication bias in the studies included in this metaanalysis. No indication of publication bias was noted from both visualization of funnel plot and the Egger’s test (P = 0.878).
N. Tang et al. / Lung Cancer 67 (2010) 17–22
21
Fig. 3. Risk estimates from studies assessing the association between an increment of coffee consumption of 2 cups/day and lung cancer risk. Squares indicated study-specific risk estimates (size of square reflects the study-statistical weight, i.e. inverse of variance); horizontal lines indicate 95% confidence intervals; diamond indicates summary relative risk estimate with its corresponding 95% confidence interval.
Q = 2.13, P = 0.144, I2 = 53.1%). However, a borderline significant inverse association was noted in non-smokers (RR = 0.78, 95% CI = 0.60–1.00; Q = 1.04, P = 0.594, I2 = 0.0%) (Table 2). Two studies have examined the association between decaffeinated coffee consumption and risk of lung cancer. Interestingly, the pooled results of these two studies found that there was a statistically significant reduction in lung cancer risk for high decaffeinated coffee consumption (RR = 0.66, 95% CI = 0.54–0.81; Q = 0.54, P = 0.461, I2 = 0.0%) (Table 2). We also examined if histological subtypes of lung cancer affected the pooled RR. However, no statistically significant association was observed in this subgroup analyses. Given the wide array of measurement categories reported in the literature, we performed a dose–response analysis and calculated a risk of lung cancer for an increase in coffee consumption of 2 cups/day (Fig. 3). Overall, we found that an increment of coffee consumption of 2 cups/day was statistically significantly associated with an 11% increased risk of developing lung cancer (RR = 1.14, 95% CI = 1.04–1.26; Q = 13.61, P = 0.093, I2 = 41.2%). 4. Discussion This meta-analysis evaluated the potential association between coffee consumption and lung cancer risk based on 5 prospective and 8 case–control studies. Overall, the summary RR indicated a significant positive association between highest coffee consumption and lung cancer risk. Furthermore, an increase in coffee consumption of 2 cups/day was statistically significantly associated with a 14% increased risk of developing lung cancer. The biologic mechanism whereby coffee increases the risk of lung cancer is likely to be multifactorial. Coffee (60–75%) is the major source of caffeine in diet. Previous studies have demonstrated that caffeine can affect DNA repair, modify the apoptotic response and perturb cell cycle checkpoint integrity [32–38]. It can also influence the normal induction of p53 in response to DNA damage through modifying p53 status [37,38]. In addition, study by Liu [39] has shown that caffeine can reduce the antioxidant and anticancer activities of flavonoids which have been suggested to play an
important role against the risk of lung cancer [40]. Furthermore, caffeine can induce the production of cytochrome P450 (CYP) enzymes such as CYP1A1/1A2 [41]. CYP1A1 was reported to activate a number of chemical carcinogens, including polycyclic hydrocarbons and their oxygenated derivatives, heterocyclic amines, aromatic amines, and nitropolycylic hydrocarbons [42]. Although the effect of coffee on lung cancer is biologically plausible, we could not rule out the possibility that the association might be confounded by smoking status. Cigarette smoking is one of the established risk factors for lung cancer. It has been reported that high intakes of coffee are frequently associated with cigarette smoking [43]. In addition, the stratified analysis by smoking status in this study indicated a borderline significant inverse association between coffee consumption and lung cancer among non-smokers but a non-significant positive association among smokers, though the non-significant positive association may be by chance because only 2 studies were included in this analysis. Therefore, our findings of a positive association between coffee consumption and lung cancer risk should be interpreted with caution. When subgroup analysis was conducted by study design, we noted that statistically significant positive association was only observed among prospective studies but not in case–control studies. This may be due to the differential misclassification of coffee consumption in case–control studies, because the information on coffee consumption was collected after the patients were diagnosed. Thus, the collected consumption data may not reflect relevant exposures considering the long latency of lung cancer. In stratified analysis by study population, we found that coffee consumption was associated with increased risk of lung cancer in studies conducted in America and Japan but not in studies conducted in Europe. The different observations may be explained at least in part, by the variations of coffee consumption or preparation methods among different countries. In addition, a possible role of ethnic differences in genetic backgrounds and the environment they lived in should also be taken into consideration. Interestingly, when we pooled the 2 studies on the association of decaffeinated coffee with lung cancer, we observed a protective effect of high decaffeinated coffee consumption on lung cancer.
22
N. Tang et al. / Lung Cancer 67 (2010) 17–22
Decaffeinated coffee contains very little caffeine, but does not reduce the levels of cafestol and kahweol. Importantly, cafestol and kahweol have been shown to produce a broad range of biochemical effects resulting in a reduction of genotoxicity of carcinogens [4]. In addition, it is also likely that the observed effect of decaffeinated coffee on lung cancer may be due to chance because only 2 studies were included in the analysis. Therefore, our results should be interpreted with caution. Some limitations of this meta-analysis should be acknowledged. First, only observational studies were included in our meta-analysis. Observational studies, even when well controlled, are susceptible to various biases. However, prospective studies, which are less susceptible to bias because of the prospective design, also showed a positive association between coffee consumption and lung cancer, suggesting the finding is not likely attributable to recall and selection bias. Second, our results are likely to be affected by some misclassification of coffee consumption. Coffee exposure is mostly assessed regarding the number of cups of coffee consumed daily, weekly or monthly. However, cup size may vary considerably. Third, we extracted the risk estimates that reflected the greatest degree of the control potential confounders, because it was hard to obtain raw data from each study for conducting standardized adjustments. Therefore, it was probably that the results based on the adjustment for different confounders were different from those based on standardized adjustments. Finally, only published studies were included in our meta-analysis. Therefore, publication bias may have occurred although no publication bias was indicated from both visualization of the funnel plot and Egger’s test. 5. Conclusions In summary, this meta-analysis of 5 prospective and 8 case–control studies involving 5347 lung cancer cases and 104,911 non-cases suggests that increased consumption of coffee may increase the risk of developing lung cancer. To provide a more definitive conclusion, further pooled analyses with more complete raw data or prospective cohort studies with larger sample size, well controlled confounding factors and longer duration of follow-up are needed in this area. Conflict of interest statement The authors promised there were not any possible conflicts of interest in this research. References [1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74–108. [2] Garcia M, Jemal A, Ward EM, Center MM, Hao Y, Siegel RL, et al. Global cancer facts & figures 2007. Atlanta, GA: American Cancer Society; 2007. [3] Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, et al. Cancer statistics, 2008. CA Cancer J Clin 2008;58:71–96. [4] Cavin C, Holzhaeuser D, Scharf G, Constable A, Huber WW, Schilter B. Cafestol and kahweol, two coffee specific diterpenes with anticarcinogenic activity. Food Chem Toxicol 2002;40:1155–63. [5] Deplanque G, Ceraline J, Mah-Becherel MC, Cazenave JP, Bergerat JP, Klein-Soyer C. Caffeine and the G2/M block override: a concept resulting from a misleading cell kinetic delay, independent of functional p53. Int J Cancer 2001;94:363–9. [6] Restrepo HE, Correa P, Haenszel W, Brinton LA, Franco A. A case–control study of tobacco-related cancers in Colombia. Bull Pan Am Health Organ 1989;23:405–13. [7] Zheng W, Doyle TJ, Kushi LH, Sellers TA, Hong CP, Folsom AR. Tea consumption and cancer incidence in a prospective cohort study of postmenopausal women. Am J Epidemiol 1996;144:175–82. [8] Sankaranarayanan R, Varghese C, Duffy SW, Padmakumary G, Day NE, Nair MK. A case–control study of diet and lung cancer in Kerala, south India. Int J Cancer 1994;58:644–9. [9] Kubik A, Zatloukal P, Tomasek L, Kriz J, Petruzelka L, Plesko I. Diet and the risk of lung cancer among women. A hospital-based case–control study. Neoplasma 2001;48:262–6.
[10] Kubik AK, Zatloukal P, Tomasek L, Pauk N, Havel L, Krepela E, et al. Dietary habits and lung cancer risk among non-smoking women. Eur J Cancer Prev 2004;13:471–80. [11] Kubik A, Zatloukal P, Tomasek L, Pauk N, Petruzelka L, Plesko I. Lung cancer risk among nonsmoking women in relation to diet and physical activity. Neoplasma 2004;51:136–43. [12] Kubik A, Zatloukal P, Tomasek L, Pauk N, Havel L, Dolezal J, et al. Interactions between smoking and other exposures associated with lung cancer risk in women: diet and physical activity. Neoplasma 2007;54:83–8. [13] Jacobsen BK, Bjelke E, Kvåle G, Heuch I. Coffee drinking, mortality, and cancer incidence: results from a Norwegian prospective study. J Natl Cancer Inst 1986;76:823–31. [14] Nomura A, Heilbrun LK, Stemmermann GN. Prospective study of coffee consumption and the risk of cancer. J Natl Cancer Inst 1986;76:587–90. [15] Stensvold I, Jacobsen BK. Coffee and cancer: a prospective study of 43,000 Norwegian men and women. Cancer Causes Control 1994;5:401–8. [16] Fu YY, Takezaki T, Tajima K. Risk factors of lung cancer—follow-up studies in Nagoya Japan. Zhonghua Liu Xing Bing Xue Za Zhi 1997;18:328–30. [17] Khan MM, Goto R, Kobayashi K, Suzumura S, Nagata Y, Sonoda T, et al. Dietary habits and cancer mortality among middle aged and older Japanese living in Hokkaido, Japan by cancer site and sex. Asian Pac J Cancer Prev 2004;5:58–65. [18] Axelsson G, Liljeqvist T, Andersson L, Bergman B, Rylander R. Dietary factors and lung cancer among men in west Sweden. Int J Epidemiol 1996;25:32–9. [19] Nyberg F, Agrenius V, Svartengren K, Svensson C, Pershagen G. Dietary factors and risk of lung cancer in never-smokers. Int J Cancer 1998;78:430–6. [20] Hu J, Mao Y, Dryer D, White K. Risk factors for lung cancer among Canadian women who have never smoked. Cancer Detect Prev 2002;26:129–38. [21] Mettlin C. Milk drinking, other beverage habits, and lung cancer risk. Int J Cancer 1989;43:608–12. [22] Mendilaharsu M, De Stefani E, Deneo-Pellegrini H, Carzoglio JC, Ronco A. Consumption of tea and coffee and the risk of lung cancer in cigarette-smoking men: a case–control study in Uruguay. Lung Cancer 1998;19:101–7. [23] Takezaki T, Hirose K, Inoue M, Hamajima N, Yatabe Y, Mitsudomi T, et al. Dietary factors and lung cancer risk in Japanese: with special reference to fish consumption and adenocarcinomas. Br J Cancer 2001;84:1199–206. [24] Baker JA, McCann SE, Reid ME, Nowell S, Beehler GP, Moysich KB. Associations between black tea and coffee consumption and risk of lung cancer among current and former smokers. Nutr Cancer 2005;52:15–21. [25] Kubik A, Zatloukal P, Tomasek L, Dolezal J, Syllabova L, Kara J, et al. A case–control study of lifestyle and lung cancer associations by histological types. Neoplasma 2008;55:192–9. [26] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–88. [27] Greenland S, Longnecker MP. Methods for trend estimation from summarized dose–response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301–9. [28] Orsini N, Bellocco R, Greenland S. Generalized least squares for trend estimation of summarized dose–response data. Stata J 2006;6:40–57. [29] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58. [30] Hedges LV, Pigott TD. The power of statistical tests in metaanalysis. Psychol Methods 2001;6:203–17. [31] Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629–34. [32] Mohr U, Emura M, Riebe-Imre M. Experimental studies on carcinogenicity and mutagenicity of caffeine. In: Garattini S, editor. Caffeine, coffee, and health. New York: Raven Press; 1993. p. 359–78. [33] Agapova LS, Ilyinskaya GV, Turovets NA, Ivanov AV, Chumakov PM, Kopnin BP. Chromosome changes caused by alterations of p53 expression. Mutat Res 1996;354:129–38. [34] DeFrank JS, Tang W, Powell SN. p53-null cells are more sensitive to ultraviolet light only in the presence of caffeine. Cancer Res 1996;56:5365–8. [35] Efferth T, Fabry U, Glatte P, Osieka R. Expression of apoptosis-related oncoproteins and modulation of apoptosis by caffeine in human leukemic cells. J Cancer Res Clin Oncol 1995;121:648–56. [36] Link Jr CJ, Evans MK, Cook JA, Muldoon R, Stevnsner T, Bohr VA. Caffeine inhibits gene-specific repair of UV-induced DNA damage in hamster cells and in human xeroderma pigmentosum group C cells. Carcinogenesis 1995;16:1149–55. [37] Müller WU, Bauch T, Wojcik A, Böcker W, Streffer C. Comet assay studies indicate that caffeine-mediated increase in radiation risk of embryos is due to inhibition of DNA repair. Mutagenesis 1996;11:57–60. [38] Ito K, Nakazato T, Miyakawa Y, Yamato K, Ikeda Y, Kizaki M. Caffeine induces G2/M arrest and apoptosis via a novel p53-dependent pathway in NB4 promyelocytic leukemia cells. J Cell Physiol 2003;196:276–83. [39] Liu RH. Health benefits of fruit and vegetables are from additive and synergistic combinations of phytochemicals. Am J Clin Nutr 2003;78:517S–20S. [40] Arts IC. A review of the epidemiological evidence on tea, flavonoids, and lung cancer. J Nutr 2008;138:1561S–6S. [41] Goasduff T, Dréano Y, Guillois B, Ménez JF, Berthou F. Induction of liver and kidney CYP1A1/1A2 by caffeine in rat. Biochem Pharmacol 1996;52:1915–9. [42] Nebert DW, Dalton TP, Okey AB, Gonzalez FJ. Role of aryl hydrocarbon receptormediated induction of the CYP1 enzymes in environmental toxicity and cancer. J Biol Chem 2004;279:23847–50. [43] Leviton A, Cowan L. A review of the literature relating caffeine consumption by women to their risk of reproductive hazards. Food Chem Toxicol 2002;40:1271–310.