Abstract
Associations of coffee and tea consumption with lung cancer risk have been inconsistent, and most lung cancer cases investigated were smokers. Included in this study were over 1.1 million participants from 17 prospective cohorts. Cox regression analyses were conducted to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Potential effect modifications by sex, smoking, race, cancer subtype and coffee type were assessed. After a median 8.6 years of follow-up, 20 280 incident lung cancer cases were identified. Compared with non-coffee and non-tea consumption, HRs (95% CIs) associated with exclusive coffee drinkers (≥2 cups/day) among current, former and never smokers were 1.30 (1.15–1.47), 1.49 (1.27–1.74) and 1.35 (1.15–1.58), respectively. Corresponding HRs for exclusive tea drinkers (≥2 cups/day) were 1.16 (1.02–1.32), 1.10 (0.92–1.32) and 1.37 (1.17–1.61). In general, the coffee and tea associations did not differ significantly by sex, race or histologic subtype. Our findings suggest that higher consumption of coffee or tea is associated with increased lung cancer risk. However, these findings should not be assumed to be causal because of the likelihood of residual confounding by smoking, including passive smoking, and change of coffee and tea consumption after study enrolment.
Introduction
Lung cancer is the leading cause of cancer death in both men and women in the United States (US).1 Consumption of coffee and tea, two widely consumed caffeine-containing beverages, has been previously linked to lung cancer risk, but evidence is not entirely consistent.2,3 Smoking, the most important cause of lung cancer, is positively correlated to coffee or tea consumption.4–8
Given the correlation between cigarette smoking and coffee or tea intake, there has been a major concern that previous studies on coffee or tea drinking with lung cancer risk might be biased by residual confounding from cigarette smoking, especially when an insufficient number of never smokers were included in these studies. Few prior studies evaluated the associations by histological subtypes of lung cancer. Furthermore, coffee and tea consumption has rarely been investigated simultaneously despite inverse correlations between the two behaviours. To overcome these limitations, we conducted a large-scale pooling study to comprehensively investigate individual and joint associations of coffee and tea consumption with lung cancer risk, including evaluations of potential effect modifications or differences by smoking status, sex, race, histologic subtype and type of coffee.
Methods
Study populations
This study included 17 cohort studies; seven from the US,9 and the other ten from China, Japan, Korea and Singapore.10–20
We excluded participants from this analysis who had no information provided on smoking status, coffee or tea consumption, or who had a history of any cancer (except non-melanoma skin cancer) prior to study recruitment. Ultimately, a total of 1 177 156 participants were included in our final study analysis. Basic descriptions for each included cohort study are summarized in Table 1.
Table 1.
Characteristics of the included cohort studies in the pooled analyses.
| Cohort | No. of subjectsa | No. of Cases | Years of Study Entry | Male | Follow-up yearsb | Age at baseline | Never smokers | Daily tea drinkerc | Average Tea Intake among drinkers (cups/month) | Daily coffee drinkerc | Average Coffee Intake among drinkers (cups/month) |
|---|---|---|---|---|---|---|---|---|---|---|---|
| % | Median | Median | % | % | Median | % | Median | ||||
| PLCO | 103 282 | 1473 | 1993–2004 | 49.1 | 7.6 | 63 | 47.6 | 18.0 | 66.6 | 61.4 | 83.3 |
| AARP | 436 140 | 7903 | 1995–1997 | 48.7 | 8.5 | 62 | 33.2 | 23.1 | 54.2 | 65.2 | 76.0 |
| SCCS | 67 002 | 602 | 2002–2009 | 40.4 | 4.4 | 50 | 36.5 | 15.2 | 30.4 | 38.1 | 30.9 |
| VITAL | 65 955 | 860 | 2000–2002 | 48.6 | 8.0 | 60 | 48.3 | 20.3 | 71.0 | 62.5 | 71.0 |
| IWHS | 33 573 | 955 | 1986–1986 | 0 | 19.7 | 61 | 66.0 | 16.7 | 30.4 | 78.0 | 78.2 |
| NHS I | 72 553 | 1870 | 1984–1984 | 100.0 | 31.7 | 42 | 44.0 | 14.4 | 47.0 | 62.7 | 49.6 |
| HPFS | 42 031 | 856 | 1986–1987 | 100.0 | 21.0 | 53 | 47.3 | 8.0 | 47.0 | 48.7 | 49.6 |
| Total (US) | 820 536 | 14 519 | 1984–2009 | 49.8 | 8.5 | 60 | 39.5 | 19.8 | 52.4 | 61.7 | 75.8 |
| SWHS | 72 558 | 771 | 1996–2000 | 0 | 14.1 | 50 | 97.2 | 10.6 | 50.0 | ---- | ---- |
| SMHS | 60 380 | 783 | 2001–2006 | 100.0 | 8.3 | 53 | 30.4 | 53.1 | 50.0 | ---- | ---- |
| SCHS | 60 233 | 1306 | 1993–1998 | 44.2 | 13.2 | 55 | 69.8 | 22.5 | 41.2 | 70.5 | 41.3 |
| KMCC | 8732 | 132 | 1993–2004 | 33.5 | 9.9 | 58 | 71.3 | 10.8 | 30.4 | 40.4 | 30.4 |
| KNCC | 6051 | 9 | 2002–2014 | 36.9 | 2.2 | 52 | 66.7 | 11.4 | 45.6 | 63.7 | 106.4 |
| Miyagi | 25 399 | 462 | 1990 | 53.1 | 15.6 | 50 | 53.3 | 82.2 | 105.0 | 44.3 | 45.0 |
| Miyagi 3P | 12 413 | 70 | 1984 | 45.5 | 7.0 | 53 | 59.0 | 86.3 | 120.0 | 30.6 | 49.6 |
| Ohsaki | 24 257 | 387 | 1995 | 50.0 | 11.2 | 59 | 56.1 | 83.5 | 120.0 | 39.8 | 45.0 |
| JPHC I | 39 129 | 760 | 1990–1992 | 47.6 | 20.6 | 50 | 60.4 | 71.4 | 112.9 | 35.0 | 45.6 |
| JPHC II | 47 468 | 1081 | 1993–1994 | 47.1 | 17.7 | 54 | 60.4 | 87.0 | 112.9 | 39.7 | 45.6 |
| Total (Asia) | 356 620 | 5761 | 1984–2014 | 46.1 | 13.3 | 53 | 63.9 | 49.4 | 100.0 | 30.0 | 45.6 |
| Total (All) | 1 177 156 | 20 280 | 1984–2014 | 49.1 | 8.6 | 58 | 48.3 | 28.8 | 70.5 | 52.1 | 75.8 |
Abbreviations: PLCO, Prostate, Lung, Colorectal and Ovarian Cancer screening Trial; AAPP, National Institute of Health AARP Diet and Health Study; SCCS, Southern Community Cohort Study; VITAL, Vitamins and Lifestyle Study; IWHS, Iowa Women’s Health Study; NHS, Nurses’ Health Study; HPFS, Health Professionals Follow-up Study; SWHS, Shanghai Women’s Health Study; SMHS, Shanghai Men’s Health Study; SCHS, Singapore Chinese Health Study; KMCC, Korean Multi-center Cancer Cohort Study; KNCC, Korean National Cancer Center Cohort; Miyagi, Miyagi Cohort Study; Miyagi 3P, Three Prefecture Cohort Study Miyagi; Ohsaki, Ohsaki National Health Insurance; JPHC, Japan Public Health Center-based prospective Study; F, Female; M, Male.
Including only participants who were eligible for the current pooled analysis.
Total time interval from 2 years after study enrollment to the date of diagnosis of any cancer or the final date of last contact (i.e., end of follow-up, date of death, or loss of follow-up).
Daily drinker was defined as ≥ 1 cup/day on average.
Outcome assessment
Incident lung cancer cases were identified in each cohort through linkages with national/regional cancer registries or self-reports confirmed by medical record review, according to the International Classification of Disease (ICD) codes. Carcinomas of bronchus and lung (ICD-9 162 or ICD-10 34.0 – 34.9) were included. Upon data availability, cases were further classified into 5 histologic types (adenocarcinoma, squamous cell carcinoma, other non-small cell carcinoma, small cell carcinoma, and all others).
Assessment of coffee and tea intake
Amounts of coffee and tea intake were assessed at cohort enrolment using food frequency questionnaires, most of which were validated against 24-hour food recall, one-week diet records, or dietary biomarkers.9,21–25 All questionnaires asked at least one question about the frequency or amount of coffee and tea consumption, except for Shanghai Women’s Health Study (SWHS) and Shanghai Men’s Health Study (SMHS), which did not collect information on coffee intake. Because of a low frequency of coffee consumption in the SWHS/SMHS source population,26,27 these two studies were excluded from analyses of coffee, and amount of coffee intake was treated as zero for tea, as well as joint coffee and tea consumption analyses. Information on caffeinated or decaffeinated coffee consumption was collected in six US studies: National Institutes of Health-American Association of Retired Persons (NIH-AARP), Iowa Women’s Health Study (IWHS), Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial, Southern Community Cohort Study (SCCS), Nurses’ Health Study 1 (NHS1) and Health Professionals Follow-Up Study (HPFS), including 699 126 participants in total. Information on type of tea consumed (green, black or oolong tea) was collected in six Asian studies: Japan Public Health Center-based prospective Study (JPHC) I&II, Miyagi Cohort Study, Ohsaki National Health Insurance study, SMHS and SWHS, comprising a total of 269 191 participants. For studies (NIH-AARP, PLCO, NHS1 and HPFS) that provided information on coffee and tea intake in amount instead of frequency per day, we converted the data to frequencies by defining 1 cup as 383 mL (approximately 13 fluid ounces), which is a typical American-sized coffee or tea drink.28 SMHS and SWHS collected tea drinking as the weight of dry tea leaves; we converted the consumption by assuming 1 cup of tea contained 5 grams of dry tea leaves.29
Statistical analysis
Data from each individual cohort were harmonized and pooled. Cox proportional hazards models were used to derive hazard ratios (HRs) and 95% confidence intervals (CIs) using aggregated data from all participating cohorts. Follow-up time was treated as the time metric. Time of study entry was calculated from 2 years after baseline enrolment to minimize the potential influence of reverse causality, and exit time was the date of diagnosis of any cancer, date of death, end or loss of follow-up, whichever came first. The proportional hazard assumption was tested by comparing the multivariable model with the interaction term between person-time and the exposure variables to the model without it. Subsequent Cox regression analyses applied models that were stratified by cohort study, smoking status and follow-up time (in 5-year categories) to account for the non-proportionality in hazard conferred by these variables. No evidence of deviation from the proportional hazard assumption was observed for other covariates.
Intakes of coffee and tea were analysed both categorically and continuously. Non-coffee drinkers and non-tea drinkers were treated as the reference group in the separate analyses of coffee and tea consumption. To address concerns that the reference group of tea drinking might include coffee drinkers and vice visa, we further conducted joint analysis, combining coffee and tea intake into a single variable (neither tea nor coffee drinking, 0< either <2 cups/day, only tea ≥2 cups/day, only coffee ≥2 cups/day, both ≥2 cups/day), with non-coffee and non-tea drinkers as the reference group. Additionally, stratified analyses were conducted to assess potential effect modifications by performing Cox regression analyses, separately, in subgroups of participants defined by smoking status, sex, race and histologic subtype. Wald Chi-Square tests were used to determine the significance levels of the interactions. Heterogeneity across histologic subtype of lung cancer or smoking status was assessed by I2 statistics.30 Possible nonlinear relationships were examined using restricted cubic spline regression analyses by smoking status.
All multivariable models were additionally adjusted for age at enrolment (years), sex (male or female), race (white, black, Asian, others), education (less than high school, high school, college graduate or higher), pack-years of cigarette smoking (continuous), alcohol intake (ethanol g/day, continuous), and body mass index (BMI) (<18.5, 18.5–24.99, 25.0–29.99, or ≥30 kg/m2). For covariates that had <3% missing within each study, missing values were assigned with the median for continuous variables, or with the most common category for categorical variables in each cohort. For those that had ≥3% missing (including education and alcohol intake), a multivariate imputation based on other covariates was used to estimate the missing value within each cohort, including year at study enrolment, history of diabetes, total energy intake, follow-up years, lung cancer status and death outcomes (SAS PROC MI procedure).
We conducted multiple sensitivity analyses including: using alternative adjustments for both duration and amount of smoking instead of pack-year, additionally adjusting for leisure-time physical activity, menopausal status, hormone therapy use, vegetable, red meat and total energy intake in the sub-population with available covariate information (all US cohorts and the SMHS and SWHS), and excluding the first 4 instead of 2 years for all participants. To evaluate the possible impact from residual confounding, we performed the analyses with additional adjustment for passive smoking and used repeated measures of coffee drinking in one contributing cohort, the NHS1. Furthermore, we conducted meta-analyses for results from analyses of individual studies. Heterogeneity across studies was evaluated by I2 statistics.30 We adopted fixed-effect models when no substantial heterogeneity was observed (I2 <0.50); otherwise, we used random-effect models in the meta-analysis.
Statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, US) or STATA 14 (STATA Corp., College Station, TX, US).
Results
Among 1 177 156 total participants, 20 280 incident primary lung cancer cases were identified after a median follow-up of 8.6 years, after excluding the first 2 years of follow-up. About 62% of US participants were daily coffee drinkers (defined as ≥1 cup/day on average) compared with 30% of Asian participants. In contrast, 20% of US participants were daily tea drinkers compared with 49% of Asian participants (Table 1).
In general, coffee drinkers were more likely to be older, white, male, smokers and alcohol drinkers, and have higher education and BMI. Similar patterns were observed for drinkers of regular and decaffeinated coffee. On the other hand, tea drinkers were generally more likely to be Asian, with lower education and BMI, and alcohol drinkers (Supplementary Tables 1–3).
A positive association was observed between coffee consumption and lung cancer risk, with an HR of 1.36 (95% CI, 1.28 to 1.44) for those who drank at least 3 cups of coffee/day on average compared with non-coffee drinkers. Corresponding HRs were 1.34 (95% CI, 1.24 to 1.46), 1.46 (95% CI, 1.32 to 1.61) and 1.20 (95% CI, 1.01 to 1.42) for current, former and never smokers, respectively. After stratifying by smoking status, neither sex nor race modified the coffee-lung cancer association, with exception to current smokers, in whom the association was stronger among females (P-interaction <.0001). Similarly, estimated HRs did not differ significantly by histologic subtypes of lung cancer, based on smoking status in each subgroup except among former smokers (Table 2). In the latter, a significant heterogeneity was noted despite that ≥ 3 cups of coffee/day was linked to an elevated OR for all subtypes of lung cancer. Restricted cubic spline analyses indicated a linear relationship between coffee intake and lung cancer risk among never smokers, but a non-linear relationship among ever, current or former smokers (Figure 1).
Table 2.
Associations between coffee intake and lung cancer risk: Stratified Analysis.
| Hazard Ratio (95%) |
|||||||
|---|---|---|---|---|---|---|---|
| Characteristics | No. of Cases | None | <1 cup/day | 1–3 cups/day | ≥3 cups/day | Per one cup/day increase | P for Interaction/Heterogeneity |
|
All participants (N=1 177 156) |
|||||||
|
| |||||||
| All | 20 280 | 1.0 | 1.04 (0.98 to 1.10) | 1.19 (1.13 to 1.25) | 1.36 (1.28 to 1.44) | 1.06 (1.05 to 1.06) | |
| Smoking status | 0.004 | ||||||
| Current smokers | 9973 | 1.0 | 1.03 (0.95 to 1.12) | 1.17 (1.09 to 1.27) | 1.34 (1.24 to 1.46) | 1.05 (1.04 to 1.06) | |
| Former smokers | 7283 | 1.0 | 1.05 (0.95 to 1.16) | 1.25 (1.14 to 1.37) | 1.46 (1.32 to 1.61) | 1.07 (1.06 to 1.08) | |
| Never smokers | 3204 | 1.0 | 1.08 (0.96 to 1.22) | 1.07 (0.96 to 1.20) | 1.20 (1.01 to 1.42) | 1.03 (1.00 to 1.06) | |
| Sex | 0.019 | ||||||
| Male | 11 304 | 1.0 | 1.08 (1.00 to 1.16) | 1.23 (1.15 to 1.32) | 1.38 (1.28 to 1.49) | 1.05 (1.04 to 1.07) | |
| Female | 8976 | 1.0 | 0.99 (0.91 to 1.08) | 1.15 (1.06 to 1.24) | 1.36 (1.25 to 1.494) | 1.06 (1.05 to 1.07) | |
| Race | 0.063 | ||||||
| White | 13 421 | 1.0 | 1.04 (0.96 to 1.13) | 1.22 (1.13 to 1.31) | 1.43 (1.32 to 1.54) | 1.06 (1.05 to 1.07) | |
| Black | 801 | 1.0 | 1.07 (0.87 to 1.32) | 1.17 (0.95 to 1.44) | 0.98 (0.66 to 1.43) | 1.03 (0.98 to 1.09) | |
| Asian | 5871 | 1.0 | 1.06 (0.97 to 1.16) | 1.16 (1.07 to 1.26) | 1.15 (1.03 to 1.29) | 1.03 (1.01 to 1.06) | |
| Histologic types | 0.004 | ||||||
| Adenocarcinoma | 4926 | 1.0 | 1.02 (0.90 to 1.16) | 1.22 (1.09 to 1.37) | 1.40 (1.23 to 1.58) | 1.06 (1.04 to 1.08) | |
| Squamous cell carcinoma | 2303 | 1.0 | 0.59 (0.78 to 1.15) | 1.21 (1.02 to 1.43) | 1.44 (1.20 to 1.73) | 1.09 (1.06 to 1.11) | |
| Other non-small cell | 2213 | 1.0 | 0.99 (0.82 to 1.20) | 1.18 (1.00 to 1.39) | 1.43 (1.19 to 1.71) | 1.07 (1.04 to 1.09) | |
| Small cell lung carcinoma | 1692 | 1.0 | 0.97 (0.77 to 1.22) | 1.22 (0.99 to 1.50) | 1.52 (1.23 to 1.89) | 1.08 (1.05 to 1.10) | |
| Unknown | 7263 | 1.0 | 1.08 (1.00 to 1.17) | 1.15 (1.07 to 1.24) | 1.30 (1.20 to 1.42) | 1.04 (1.03 to 1.06) | |
|
Current smokers (N=220,866) |
|||||||
|
| |||||||
| Sex | <.0001 | ||||||
| Male | 5736 | 1.0 | 1.06 (0.96 to 1.18) | 1.16 (1.06 to 1.28) | 1.28 (1.16 to 1.42) | 1.04 (1.02 to 1.05) | |
| Female | 4237 | 1.0 | 0.99 (0.86 to 1.15) | 1.23 (1.08 to 1.40) | 1.48 (1.29 to 1.69) | 1.07 (1.05 to 1.09) | |
| Race | 0.384 | ||||||
| White | 6200 | 1.0 | 0.98 (0.85 to 1.12) | 1.18 (1.05 to 1.34) | 1.40 (1.23 to 1.58) | 1.05 (1.04 to 1.06) | |
| Black | 511 | 1.0 | 1.18 (0.90 to 1.55) | 1.25 (0.95 to 1.63) | 1.18 (0.76 to 1.85) | 0.99 (0.88 to 1.10) | |
| Asian | 3169 | 1.0 | 1.08 (0.96 to 1.22) | 1.17 (1.05 to 1.31) | 1.18 (1.03 to 1.36) | 1.03 (1.00 to 1.06) | |
| Histologic types | 0.223 | ||||||
| Adenocarcinoma | 1795 | 1.0 | 0.94 (0.75 to 1.18) | 1.12 (0.92 to 1.36) | 1.42 (1.15 to 1.74) | 1.07 (1.04 to 1.09) | |
| Squamous cell carcinoma | 1232 | 1.0 | 0.82 (0.62 to 1.08) | 1.08 (0.86 to 1.36) | 1.21 (0.95 to 1.55) | 1.07 (1.03 to 1.10) | |
| Other non-small cell | 969 | 1.0 | 1.14 (0.84 to 1.55) | 1.14 (0.87 to 1.50) | 1.40 (1.05 to 1.86) | 1.05 (1.01 to 1.08) | |
| Small cell lung carcinoma | 1063 | 1.0 | 1.04 (0.76 to 1.43) | 1.30 (0.99 to 1.71) | 1.50 (1.12 to 1.99) | 1.06 (1.03 to 1.10) | |
| Unknown | 4052 | 1.0 | 1.09 (0.97 to 1.21) | 1.20 (1.08 to 1.33) | 1.37 (1.23 to 1.54) | 1.04 (1.03 to 1.06) | |
|
Former smokers (N=404,264) |
|||||||
|
| |||||||
| Sex | |||||||
| Male | 4856 | 1.0 | 1.13 (1.00 to 1.28) | 1.35 (1.20 to 1.52) | 1.59 (1.40 to 1.81) | 1.08 (1.06 to 1.10) | 0.222 |
| Female | 2427 | 1.0 | 0.92 (0.79 to 1.08) | 1.11 (0.96 to 1.29) | 1.25 (1.06 to 1.48) | 1.05 (1.03 to 1.08) | |
| Race | 0.253 | ||||||
| White | 6251 | 1.0 | 1.02 (0.91 to 1.15) | 1.22 (1.10 to 1.36) | 1.43 (1.27 to 1.60) | 1.07 (1.06 to 1.08) | |
| Black | 239 | 1.0 | 0.93 (0.64 to 1.34) | 1.03 (0.71 to 1.49) | 0.62 (0.26 to 1.48) | 1.01 (0.90 to 1.12) | |
| Asian | 714 | 1.0 | 1.14 (0.90 to 1.44) | 1.48 (1.19 to 1.85) | 1.40 (0.99 to 2.00) | 1.00 (0.95 to 1.04) | |
| Histologic types | 0.004 | ||||||
| Adenocarcinoma | 2113 | 1.0 | 1.10 (0.90 to 1.34) | 1.38 (1.15 to 1.65) | 1.46 (1.20 to 1.79) | 1.06 (1.04 to 1.09) | |
| Squamous cell carcinoma | 986 | 1.0 | 1.09 (0.81 to 1.47) | 1.36 (1.04 to 1.78) | 1.83 (1.37 to 2.44) | 1.12 (1.08 to 1.16) | |
| Other non-small cell | 1023 | 1.0 | 0.97 (0.73 to 1.28) | 1.21 (0.94 to 1.55) | 1.52 (1.16 to 1.99) | 1.09 (1.05 to 1.13) | |
| Small cell lung carcinoma | 568 | 1.0 | 1.03 (0.70 to 1.52) | 1.21 (0.85 to 1.72) | 1.86 (1.28 to 2.70) | 1.13 (1.08 to 1.18) | |
| Unknown | 1943 | 1.0 | 1.02 (0.87 to 1.20) | 1.16 (1.00 to 1.35) | 1.23 (1.03 to 1.47) | 1.05 (1.02 to 1.07) | |
|
Never smokers (N=552 026) |
|||||||
|
| |||||||
| Sex | 0.123 | ||||||
| Male | 712 | 1.0 | 1.09 (0.87 to 1.37) | 1.18 (0.95 to 1.47) | 0.98 (0.68 to 1.41) | 1.00 (0.94 to 1.06) | |
| Female | 2312 | 1.0 | 1.08 (0.94 to 1.24) | 1.03 (0.91 to 1.18) | 1.26 (1.03 to 1.54) | 1.04 (1.00 to 1.08) | |
| Race | 0.870 | ||||||
| White | 970 | 1.0 | 1.25 (1.02 to 1.54) | 1.17 (0.96 to 1.42) | 1.37 (1.07 to 1.76) | 1.04 (1.00 to 1.08) | |
| Black | 51 | 1.0 | 0.89 (0.45 to 1.78) | 1.30 (0.63 to 2.67) | ------- | 1.10 (0.81 to 1.50) | |
| Asian | 1988 | 1.0 | 1.00 (0.86 to 1.17) | 1.03 (0.89 to 1.20) | 1.05 (0.79 to 1.38) | 1.01 (0.96 to 1.07) | |
| Histologic types | 0.770 | ||||||
| Adenocarcinoma | 1018 | 1.0 | 1.06 (0.83 to 1.36) | 1.11 (0.90 to 1.37) | 1.12 (0.80 to 1.56) | 1.03 (0.97 to 1.08) | |
| Squamous cell carcinoma | 85 | 1.0 | 1.21 (0559 to 2.369) | 1.33 (0.68 to 2.58) | 1.44 (0.51 to 4.09) | 1.10 (0.92 to 1.31) | |
| Other non-small cell | 221 | 1.0 | 0.78 (0.48 to 1.27) | 1.22 (0.83 to 1.80) | 1.25 (0.70 to 2.23) | 1.07 (0.97 to 1.18) | |
| Small cell lung carcinoma | 61 | 1.0 | 0.38 (0.15 to 1.01) | 1.00 (0.52 to 1.94) | 1.03 (0.35 to 2.99) | 1.13 (0.89 to 1.42) | |
| Unknown | 1268 | 1.0 | 1.15 (0.99 to 1.34) | 1.00 (0.85 to 1.17) | 1.20 (0.94 to 1.53) | 1.02 (0.97 to 1.07) | |
HRs were stratified by cohort, smoking status and enrollment year (in 5-year intervals).
HRs were additionally adjusted for age at enrollment, sex, race, education, packs-years of cigarette smoking, alcohol intake, tea intake and BMI.
P for Interaction/Heterogeneity were assessed by comparing the HRs of per one cup/day increase between groups.
Figure 1.
Dose-response relationship between coffee intake and lung cancer risk according to smoking status. Solid line indicates hazard ratio of lung cancer and dashed line indicates the 95% confidence intervals. Non-coffee and non-tea drinkers were set as the reference group for the analysis of coffee and tea, respectively. Ever smokers were consisted of former smokers and current smokers. For plots of coffee analyses, knots were placed at 5th, 25th, 50th, 75th and 95th percentiles for ever and former smokers, and at 5th, 50th and 95th percentiles for never and current smokers. For plots of tea analyses, knots were posited at 5th, 25th, 50th, 75th and 95th percentiles for ever, current and former smokers, and at 5th, 50th and 95th percentiles for never smokers. All models were stratified by cohort and enrolment year with additional adjustment for age at enrolment, sex, race, education, packs-years of cigarette smoking, alcohol intake, body mass index, and coffee or tea intake.
In an analysis of 699 126 participants with information on type of coffee consumption, we found that regular or decaffeinated coffee drinking showed similar associations with lung cancer risk after multivariable and mutual adjustments, with respective HRs of 1.30 (95% CI, 1.23 to 1.37) and 1.27 (95% CI, 1.19 to 1.35) for drinking two or more cups per day versus not drinking. A positive association was observed among never smokers (HR=1.33, 95% CI, 1.06 to 1.66) for decaffeinated coffee consumption. However, the association shown for regular coffee intake in never smokers was not significant (HR=1.10, 95%CI 0.91–1.32) (Table 3).
Table 3.
Association of coffee consumption with lung cancer risk according to type of coffee consumed.
| Hazard Ratio (95%) |
|||||||
|---|---|---|---|---|---|---|---|
| Coffee type | Total No. | No. of cases | All (N=699 126) | Current smoker (N=114 658) | Former smokers (N=318 518) | Never smokers (N=265 950) | P for Heterogeneity |
| Regular | |||||||
| None | 253 987 | 3529 | 1.0 | 1.0 | 1.0 | 1.0 | |
| <1 cup/day | 127 566 | 1764 | 1.02 (0.95 to 1.08) | 0.97 (0.88 to 1.07) | 1.03 (0.94 to 1.14) | 1.07 (0.90 to 1.27) | |
| 1–2 cups/day | 80 825 | 1471 | 1.08 (1.01 to 1.16) | 1.09 (0.98 to 1.21) | 1.06 (0.95 to 1.18) | 1.09 (0.87 to 1.36) | |
| ≥2 cups/day | 236 748 | 5990 | 1.30 (1.23 to 1.37) | 1.31 (1.21 to 1.42) | 1.30 (1.19 to 1.41) | 1.10 (0.91 to 1.32) | 0.226 |
| Per one cup/day increase | 1.07 (1.05 to 1.08) | 1.06 (1.05 to 1.08) | 1.08 (1.06 to 1.10) | 1.04 (0.99 to 1.10) | |||
| Decaffeinated | |||||||
| None | 443 650 | 8417 | 1.0 | 1.0 | 1.0 | 1.0 | |
| <1 cup/day | 130 346 | 1813 | 1.01 (0.95 to 1.07) | 1.00 (0.92 to 1.09) | 0.98 (0.90 to 1.08) | 1.17 (0.99 to 1.38) | |
| 1–2 cups/day | 41 339 | 695 | 1.09 (0.95 to 1.07) | 1.12 (1.00 to 1.26) | 1.06 (0.92 to 1.21) | 1.00 (0.78 to 1.29) | |
| ≥2 cups/day | 84 791 | 1829 | 1.27 (1.19 to 1.35) | 1.16 (1.05 to 1.28) | 1.34 (1.21 to 1.47) | 1.33 (1.06 to 1.66) | 0.109 |
| Per one cup/day increase | 1.07 (1.05 to 1.09) | 1.04 (1.01 to 1.06) | 1.10 (1.08 to 1.13) | 1.08 (1.01 to 1.15) | |||
HRs were stratified by cohort, smoking status and enrollment year (in 5-year intervals) and were additionally adjusted for age at enrollment, sex, race, education, packs-years of cigarette smoking, alcohol intake, tea intake, BMI and each other.
P for Interaction/Heterogeneity were assessed by comparing the HRs of per one cup/day increase between groups.
The association between tea intake and lung cancer risk differed by smoking status. Among current smokers, tea intake was not associated with lung cancer risk; among former smokers, higher consumption of tea was inversely associated with the risk (HR=0.85, 95% CI, 0.75 to 0.97); in never smokers, higher tea intake was associated with an increased risk (HR=1.28, 95% CI, 1.10 to 1.50; P for heterogeneity = 0.004). In general, associations between tea and lung cancer did not vary by sex, race or histologic subtype after stratifying by smoking status. One exception was the significant interaction between race and tea drinking observed among current smokers (P-interaction = 0.01). A significant tea-lung cancer association was only observed in Asian populations (HR=1.13, 95% CI, 1.01 to 1.27) (Table 4). Restricted cubic spline regression analyses suggested a linear relationship between tea consumption and lung cancer risk among never smokers, and the overall association was not statistically significant for tea drinking among current smokers (P = 0.47) (Figure 1).
Table 4.
Associations between tea intake and lung cancer risk: Stratified Analysis.
| Hazard Ratio (95%) |
|||||||
|---|---|---|---|---|---|---|---|
| Characteristics | No. of Cases | None | <1 cup/day | 1–3 cups/day | ≥3 cups/day | Per one cup/day increase | P for Interaction/ Heterogeneity |
|
All participants (N=1 177 156) |
|||||||
|
| |||||||
| All | 20 280 | 1.0 | 0.99 (0.96 to 1.03) | 0.99 (0.94 to 1.04) | 1.04 (0.98 to 1.11) | 1.01 (0.99 to 1.02) | |
| Smoking status | 0.004 | ||||||
| Current smokers | 9973 | 1.0 | 0.99 (0.94 to 1.05) | 0.98 (0.92 to 1.05) | 1.08 (0.99 to 1.17) | 1.01 (0.99 to 1.02) | |
| Former smokers | 7283 | 1.0 | 0.91 (0.85 to 0.97) | 0.91 (0.84 to 0.99) | 0.85 (0.75 to 0.97) | 0.98 (0.96 to 1.01) | |
| Never smokers | 3263 | 1.0 | 1.20 (1.09 to 1.33) | 1.18 (1.05 to 1.33) | 1.28 (1.10 to 1.50) | 1.04 (1.02 to 1.07) | |
| Sex | 0.215 | ||||||
| Male | 11 304 | 1.0 | 0.98 (0.93 to 1.03) | 0.94 (0.88 to 1.00) | 1.02 (0.94 to 1.10) | 1.00 (0.99 to 1.02) | |
| Female | 8976 | 1.0 | 1.01 (0.95 to 1.06) | 1.04 (0.97 to 1.10) | 1.04 (0.94 to 1.16) | 1.01 (0.99 to 1.03) | |
| Race | 0.002 | ||||||
| White | 13 421 | 1.0 | 0.95 (0.91 to 0.99) | 0.94 (0.88 to 1.00) | 0.90 (0.81 to 1.00) | 0.98 (0.96 to 1.00) | |
| Black | 801 | 1.0 | 1.11 (0.93 to 1.34) | 1.13 (0.88 to 1.47) | 0.99 (0.51 to 1.89) | 1.01 (0.93 to 1.10) | |
| Asian | 5871 | 1.0 | 1.09 (1.00 to 1.18) | 1.08 (1.00 to 1.18) | 1.16 (1.07 to 1.27) | 1.02 (1.01 to 1.04) | |
| Histologic types | 0.458 | ||||||
| Adenocarcinoma | 4926 | 1.0 | 1.05 (0.97 to 1.13) | 1.04 (0.95 to 1.15) | 1.02 (0.87 to 1.19) | 1.00 (0.97 to 1.02) | |
| Squamous cell carcinoma | 2303 | 1.0 | 0.97 (0.87 to 1.08) | 1.00 (0.88 to 1.14) | 0.93 (0.74 to 1.17) | 0.99 (0.95 to 1.03) | |
| Other non-small cell | 2213 | 1.0 | 0.97 (0.87 to 1.09) | 0.88 (0.77 to 1.02) | 0.95 (0.74 to 1.21) | 0.97 (0.93 to 1.02) | |
| Small cell lung carcinoma | 1692 | 1.0 | 0.96 (0.85 to 1.10) | 0.97 (0.83 to 1.14) | 0.96 (0.73 to 1.25) | 0.99 (0.95 to 1.04) | |
| Unknown | 7263 | 1.0 | 0.96 (0.91 to 1.03) | 0.94 (0.86 to 1.02) | 1.05 (0.96 to 1.15) | 1.01 (1.00 to 1.03) | |
|
Current smokers (N=220 866) |
|||||||
|
| |||||||
| Sex | 0.004 | ||||||
| Male | 5736 | 1.0 | 0.99 (0.92 to 1.07) | 0.96 (0.88 to 1.04) | 1.07 (0.97 to 1.18) | 1.01 (1.00 to 1.03) | |
| Female | 4237 | 1.0 | 0.99 (0.92 to 1.06) | 1.02 (0.92 to 1.13) | 1.04 (0.88 to 1.24) | 1.00 (0.97 to 1.03) | |
| Race | 0.014 | ||||||
| White | 6200 | 1.0 | 0.96 (0.91 to 1.03) | 0.95 (0.87 to 1.07) | 0.91 (0.77 to 1.07) | 0.98 (0.95 to 1.00) | |
| Black | 511 | 1.0 | 1.04 (0.83 to 1.31) | 1.15 (0.84 to 1.57) | 0.97 (0.42 to 2.24) | 0.99 (0.88 to 1.10) | |
| Asian | 3169 | 1.0 | 1.05 (0.93 to 1.18) | 1.02 (0.91 to 1.14) | 1.13 (1.01 to 1.27) | 1.02 (1.00 to 1.04) | |
| Histologic types | 0.848 | ||||||
| Adenocarcinoma | 1795 | 1.0 | 0.99 (0.88 to 1.12) | 1.00 (0.86 to 1.16) | 1.10 (0.88 to 1.37) | 1.02 (0.98 to 1.06) | |
| Squamous cell carcinoma | 1232 | 1.0 | 0.97 (0.84 to 1.12) | 0.95 (0.80 to 1.14) | 1.00 (0.75 to 1.33) | 0.99 (0.94 to 1.04) | |
| Other non-small cell | 969 | 1.0 | 0.97 (0.83 to 1.15) | 0.92 (0.75 to 1.14) | 1.02 (0.72 to 1.45) | 0.99 (0.93 to 1.05) | |
| Small cell lung carcinoma | 1063 | 1.0 | 1.02 (0.87 to 1.19) | 1.00 (0.82 to 1.22) | 1.19 (0.87 to 1.62) | 1.02 (0.96 to 1.07) | |
| Unknown | 4052 | 1.0 | 0.97 (0.90 to 1.06) | 0.94 (0.84 to 1.05) | 1.08 (0.96 to 1.21) | 1.01 (0.99 to 1.03) | |
|
Former smokers (N=404 264) |
|||||||
|
| |||||||
| Sex | 0.797 | ||||||
| Male | 4856 | 1.0 | 0.91 (0.85 to 0.99) | 0.90 (0.82 to 1.00) | 0.87 (0.75 to 1.02) | 0.99 (0.96 to 1.01) | |
| Female | 2427 | 1.0 | 0.91 (0.82 to 1.02) | 0.94 (0.82 to 1.08) | 0.80 (0.63 to 1.01) | 0.98 (0.94 to 1.02) | |
| Race | 0.900 | ||||||
| White | 6251 | 1.0 | 0.89 (0.84 to 0.96) | 0.89 (0.82 to 0.97) | 0.81 (0.69 to 0.95) | 0.98 (0.95 to 1.01) | |
| Black | 239 | 1.0 | 1.20 (0.83 to 1.72) | 1.16 (0.71 to 1.88) | 0.84 (0.25 to 2.74) | 1.04 (0.89 to 1.22) | |
| Asian | 714 | 1.0 | 0.90 (0.71 to 1.14) | 0.99 (0.78 to 1.24) | 0.97 (0.75 to 1.25) | 1.00 (0.95 to 1.04) | |
| Histologic types | 0.571 | ||||||
| Adenocarcinoma | 2113 | 1.0 | 0.92 (0.81 to 1.04) | 0.94 (0.81 to 1.09) | 0.83 (0.64 to 1.08) | 0.97 (0.93 to 1.02) | |
| Squamous cell carcinoma | 986 | 1.0 | 0.96 (0.80 to 1.14) | 1.04 (0.84 to 1.29) | 0.79 (0.52 to 1.20) | 1.00 (0.94 to 1.07) | |
| Other non-small cell | 1023 | 1.0 | 0.98 (0.82 to 1.16) | 0.81 (0.65 to 1.01) | 0.80 (0.53 to 1.19) | 0.94 (0.87 to 1.00) | |
| Small cell lung carcinoma | 568 | 1.0 | 0.89 (0.71 to 1.11) | 0.88 (0.66 to 1.17) | 0.46 (0.24 to 0.88) | 0.94 (0.85 to 1.03) | |
| Unknown | 1943 | 1.0 | 0.86 (0.77 to 0.96) | 0.86 (0.73 to 1.01) | 0.84 (0.69 to 1.04) | 0.99 (0.95 to 1.02) | |
|
Never smokers (N=552 026) |
|||||||
|
| |||||||
| Sex | 0.501 | ||||||
| Male | 712 | 1.0 | 1.33 (1.09 to 1.62) | 1.17 (0.92 to 1.48) | 1.29 (0.94 to 1.76) | 1.04 (0.98 to 1.10) | |
| Female | 2312 | 1.0 | 1.16 (1.03 to 1.29) | 1.18 (1.03 to 1.35) | 1.27 (1.06 to 1.52) | 1.04 (1.01 to 1.08) | |
| Race | 0.976 | ||||||
| White | 970 | 1.0 | 1.25 (1.06 to 1.48) | 1.19 (0.96 to 1.47) | 1.57 (1.09 to 2.26) | 1.03 (0.97 to 1.10) | |
| Black | 51 | 1.0 | 1.50 (0.68 to 3.30) | 0.87 (0.28 to 2.75) | 3.17 (0.37 to 27.20) | 1.15 (0792 to 1.67) | |
| Asian | 1988 | 1.0 | 1.17 (1.03 to 1.32) | 1.17 (1.02 to 1.36) | 1.22 (1.03 to 1.45) | 1.04 (1.01 to 1.08) | |
| Histologic types | 0.797 | ||||||
| Adenocarcinoma | 1018 | 1.0 | 1.35 (1.16 to 1.58) | 1.25 (1.04 to 1.51) | 1.14 (0.76 to 1.70) | 1.01 (0.94 to 1.08) | |
| Squamous cell carcinoma | 85 | 1.0 | 1.06 (0.63 to 1.79) | 1.21 (0.65 to 2.25) | 1.17 (0.28 to 5.01) | 1.07 (0.86 to 1.34) | |
| Other non-small cell | 221 | 1.0 | 0.89 (0.63 to 1.26) | 1.12 (0.76 to 1.66) | 1.32 (0.65 to 2.70) | 1.08 (0.95 to 1.22) | |
| Small cell lung carcinoma | 61 | 1.0 | 0.67 (0.35 to 1.32) | 1.40 (0.70 to 2.77) | 1.60 (0.46 to 5.56) | 1.12 (0.89 to 1.42) | |
| Unknown | 1268 | 1.0 | 1.21 (1.02 to 1.44) | 1.12 (0.91 to 1.38) | 1.28 (1.04 to 1.58) | 1.05 (1.02 to 1.08) | |
HRs were stratified by cohort, smoking status and enrollment year (in 5-year intervals).
HRs were additionally adjusted for age at enrollment, sex, race, education, packs-years of cigarette smoking, alcohol intake, coffee intake and BMI.
P for Interaction/Heterogeneity were assessed by comparing the HRs of per one cup/day increase between groups.
Analyses by tea type were conducted among six Asian cohorts. We observed relatively consistent positive associations for green tea. Compared with non-green tea drinkers, those who drank at least 2 cups/day of green tea had a higher likelihood of developing lung cancer for either current smokers (HR=1.22, 95% CI, 1.07–1.38) or never smokers (HR=1.31, 95% CI, 1.10–1.57). Oolong tea was found to be associated with an increased risk of lung cancer among never smokers (HR=1.43, 95% CI, 1.07–1.91). There were no significant associations observed for black tea (Supplementary Table 4).
Joint analyses of coffee and tea consumption, with individuals who drank neither coffee nor tea as the reference group, showed positive lung cancer associations with both coffee and tea consumption. HRs for lung cancer associated with exclusive coffee drinking (≥2 cups/day) among current, former and never smokers were 1.30 (95% CI, 1.15 to 1.47), 1.49 (95% CI, 1.27 to 1.74) and 1.35 (95% CI, 1.15 to 1.58), respectively. Corresponding HRs associated with exclusive tea drinking (≥2 cups/day) were 1.16 (95% CI, 1.02 to 1.32), 1.10 (95% CI, 0.92 to 1.32) and 1.37 (95% CI, 1.17 to 1.61). After stratifying by smoking status, a significant heterogeneity by sex was observed among current smokers (P-interaction < 0.001). The association for exclusive coffee drinkers was stronger for females (HR=1.38, 95% CI, 1.11 to 1.71) than males (HR=1.27, 95% CI, 1.09 to 1.47), and the association for exclusive tea drinkers was only significant among males (HR=1.18, 95% CI, 1.02 to 1.37) (Table 5).
Table 5.
Associations of tea and coffee intake with lung cancer risk according to joint distributions of coffee and tea intake: Stratified Analysis.
| Hazard Ratio (95%) |
|||||||
|---|---|---|---|---|---|---|---|
| Characteristics | No. of Cases | Neither | Either (0–2 cup/day) | Only Tea (≥2 cups/day) | Only Coffee (≥2 cups/day) | Both (≥2 cusp/day) | P for Interaction/ Heterogeneity |
|
All participants (N=1 177 156) |
|||||||
|
| |||||||
| All | 20 280 | 1.0 | 1.19 (1.10 to 1.28) | 1.24 (1.14 to 1.35) | 1.46 (1.35 to 1.58) | 1.37 (1.24 to 1.52) | |
| Smoking status | 0.326 | ||||||
| Current smokers | 9973 | 1.0 | 1.04 (0.93 to 1.17) | 1.16 (1.02 to 1.32) | 1.30 (1.15 to 1.47) | 1.21 (1.05 to 1.41) | |
| Former smokers | 7283 | 1.0 | 1.17 (1.00 to 1.37) | 1.10 (0.92 to 1.32) | 1.49 (1.27 to 1.74) | 1.39 (1.15 to 1.69) | |
| Never smokers | 3204 | 1.0 | 1.33 (1.18 to 1.50) | 1.37 (1.17 to 1.61) | 1.35 (1.15 to 1.58) | 1.48 (1.13 to 1.95) | |
| Sex | 0.049 | ||||||
| Male | 11 304 | 1.0 | 1.15 (1.04 to 1.28) | 1.22 (1.09 to 1.37) | 1.44 (1.29 to 1.60) | 1.28 (1.11 to 1.47) | |
| Female | 8976 | 1.0 | 1.21 (1.09 to 1.35) | 1.23 (1.07 to 1.40) | 1.49 (1.33 to 1.67) | 1.50 (1.29 to 1.75) | |
| Race | 0.031 | ||||||
| White | 13 421 | 1.0 | 1.18 (1.03 to 1.35) | 1.12 (0.96 to 1.32) | 1.47 (1.29 to 1.68) | 1.39 (1.19 to 1.63) | |
| Black | 801 | 1.0 | 1.08 (0.79 to 1.47) | 1.12 (0.67 to 1.87) | 1.17 (0.83 to 1.65) | 1.01 (0.56 to 1.83) | |
| Asian | 5874 | 1.0 | 1.22 (1.10 to 1.34) | 1.30 (1.17 to 1.45) | 1.37 (1.21 to 1.55) | 1.31 (1.12 to 1.54) | |
|
Current smokers (N=220 866) |
|||||||
|
| |||||||
| Sex | <.001 | ||||||
| Male | 5736 | 1.0 | 1.04 (0.90 to 1.20) | 1.18 (1.02 to 1.37) | 1.27 (1.09 to 1.47) | 1.13 (0.94 to 1.35) | |
| Female | 4237 | 1.0 | 1.07 (0.86 to 1.34) | 1.06 (0.82 to 1.37) | 1.38 (1.11 to 1.71) | 1.43 (1.10 to 1.84) | |
| Race | 0.420 | ||||||
| White | 6200 | 1.0 | 0.98 (0.79 to 1.21) | 1.00 (0.78 to 1.29) | 1.25 (1.01 to 1.54) | 1.13 (0.89 to 1.44) | |
| Black | 511 | 1.0 | 1.04 (0.71 to 1.53) | 0.68 (0.31 to 1.49) | 1.15 (0.76 to 1.76) | 1.18 (0.59 to 2.35) | |
| Asian | 3169 | 1.0 | 1.09 (0.94 to 1.28) | 1.22 (1.04 to 1.42) | 1.26 (1.05 to 1.51) | 1.27 (1.03 to 1.56) | |
|
Former smokers (N=404 264) |
|||||||
|
| |||||||
| Sex | 0.976 | ||||||
| Male | 4856 | 1.0 | 1.16 (0.96 to 1.40) | 1.08 (0.87 to 1.34) | 1.51 (1.25 to 1.83) | 1.42 (1.12 to 1.80) | |
| Female | 2427 | 1.0 | 1.21 (0.90 to 1.61) | 1.15 (0.83 to 1.60) | 1.47 (1.10 to 1.97) | 1.40 (0.99 to 1.97) | |
| Race | 0.296 | ||||||
| White | 6251 | 1.0 | 1.18 (0.97 to 1.44) | 1.11 (0.88 to 1.40) | 1.51 (1.24 to 1.84) | 1.45 (1.15 to 1.83) | |
| Black | 239 | 1.0 | 1.26 (0.66 to 2.40) | 1.79 (0.76 to 4.22) | 1.30 (0.65 to 2.60) | 0.81 (0.22 to 2.97) | |
| Asian | 714 | 1.0 | 1.14 (0.85 to 1.52) | 1.06 (0.78 to 1.45) | 1.38 (0.96 to 1.97) | 1.00 (0.58 to 1.71) | |
|
Never smokers (N=552 026) |
|||||||
|
| |||||||
| Sex | 0.356 | ||||||
| Male | 712 | 1.0 | 1.66 (1.27 to 2.17) | 1.53 (1.09 to 2.16) | 1.56 (1.13 to 2.16) | 1.55 (0.87 to 2.76) | |
| Female | 2312 | 1.0 | 1.25 (1.09 to 1.43) | 1.33 (1.10 to 1.60) | 1.30 (1.09 to 1.56) | 1.45 (1.06 to 1.97) | |
| Race | 0.751 | ||||||
| White | 970 | 1.0 | 1.72 (1.24 to 2.39) | 1.45 (0.95 to 2.23) | 1.64 (1.18 to 2.30) | 2.20 (1.38 to 3.50) | |
| Black | 51 | 1.0 | 0.89 (0.34 to 2.32) | 1.49 (0.34 to 6.51) | 1.02 (0.29 to 3.62) | -------- | |
| Asian | 1988 | 1.0 | 1.27 (1.12 to 1.45) | 1.35 (1.13 to 1.62) | 1.37 (1.09 to 1.72) | 1.24 (0.85 to 1.80) | |
HRs were stratified by cohort, smoking status and enrollment year (in 5-year intervals) and were additionally adjusted for age at enrollment, sex, race, education, packs-years of cigarette smoking, alcohol intake and BMI.
Results were robust in sensitivity analyses with a different approach to adjusting for smoking, controlling for additional covariates and further exclusion of follow-up time (data not shown). In general, meta-analyses yielded similar results to those of aggregated analyses in terms of all coffee, regular coffee, decaffeinated coffee and tea intake. However, the association between coffee drinking and lung cancer risk did not reach statistical significance among never smokers (HR=1.03, 95% CI, 0.99 to 1.06) in the meta-analysis. Forest plots showing individual study results by treating exposures as continuous variables are presented in Supplementary Figures 1–4.
Discussion
In this pooled analysis of near 1.2 million participants, we observed that frequent coffee (either decaffeinated or regular coffee) and tea consumption were both associated with an increased lung cancer risk. Positive association patterns were seen consistently across race and smoking status. Observed associations did not differ markedly by histologic subtype.
Our finding of a positive association between coffee consumption and lung cancer risk is consistent with a recent meta-analysis.2 However, our positive finding for tea consumption was inconsistent with a meta-analysis of prospective cohort studies.3 Coffee and tea are both caffeine-rich beverages. Caffeine shares the CYP1A2 metabolizing pathway with polycyclic aromatic hydrocarbons, carcinogens from tobacco smoking. As inducers for cytochrome P450, both caffeine and polycyclic aromatic hydrocarbons from tobacco smoking can upregulate the CYP1A2 enzyme, which may result in increased caffeine tolerance and consumption.31–33 Coffee and tea drinking are often consumed with smoking,34 and confounding from smoking has been the prevalent explanation for observed positive coffee- and tea-lung cancer association, with particular concern on insufficient adjustments for amount of smoking or second-hand smoking. However, a recent report from the Lung Cancer Cohort Consortium indicated that self-reported smoking alone should be adequate for the adjustment of smoking; adding circulating cotinine levels to models that included self-reported smoking variables only provided limited improvement for predicting lung cancer risk among current smokers. Moreover, that study found that the percentage of participants with cotinine concentrations indicating active smoking was very low among never smokers (2.7% for cases and 0.8% for controls).35 Furthermore, no significant association was observed between cotinine level indicative of passive smoking and lung cancer risk.35 We conducted sensitivity analyses in the SWHS to additionally adjust for self-reported passive smoking from the working environment and family members, and found no appreciable changes in the tea-lung cancer association (data not shown). Additionally, the observed coffee- and tea-lung cancer associations did not vary between Asians (passive smoking is more common) and Americans, which supports that residual confounding from passive smoking may not be the only explanation of our findings. Finally, we observed inverse associations of coffee and tea drinking with total mortality in our study population, which also does not support residual confounding due to smoking as the sole explanation for observed positive associations (data not shown). On the other hand, in the NHS1, where we were able to additionally adjust for passive smoking and use updated information on coffee intake and smoking behaviour beyond baseline evaluation, associations between coffee intake and lung cancer risk were attenuated after adjustment for passive smoking and became non-significant when consumption information during follow-up was taken into consideration (Supplementary Tables 5 and 6). It is possible that smokers at baseline may have quit smoking and reduced their coffee intakes during the follow-up period, while smokers who quit before baseline may have been more likely to keep their coffee consumption levels during the follow-up period. If this speculation were true, it may explain the slightly stronger coffee and lung cancer association that we observed among baseline former smokers than baseline current smokers. It may also explain the attenuation of the association observed in the NHS1. However, due to the lack of passive smoking data and longitudinal measurements of coffee or tea drinking and smoking habits from the vast majority of contributing cohorts, similar analyses could not be performed in other cohorts. Therefore, we cannot assume that we have fully removed the effects of residual confounding due to incomplete adjustments for changes in active and passive smoking or drinking habits after study enrolment. Furthermore, we found that the coffee-lung cancer association was stronger in women than men among current smokers. This difference is likely explainable by residual confounding due to smoking given the fact that women are more reluctant to disclose their smoking behaviours, especially in cultures like Asian countries.36
It has been previously reported that the coffee roasting process can result in significant amounts of carcinogenic small molecules such as acrylamide,37 which has been shown to be mutagenic in mouse lungs.38 Interestingly, in a subgroup of participants with information on type of coffee drinking, we only found a significant positive association for decaffeinated coffee. However, decaffeinated coffee drinkers are likely to have consumed regular coffee.39 Thus, residual effects from caffeine cannot be excluded.
The positive association for tea was unexpected. One speculative explanation might be the contamination of pesticides, which are ubiquitously used in tea cultivation40 and have been suggested to be linked to a higher lung cancer risk. Additionally, both coffee and tea have been found to contain gallic acid and pyrogallol, which are responsible for potent DNA-damaging activity at concentrations consumed dietarily.41 Inducing cytochrome P450 could be an alternative speculation.
To our knowledge, this is the largest prospective study conducted to date that examined the associations of coffee and tea intake with lung cancer risk. Since smoking is the most significant risk factor for lung cancer and is closely related to coffee and tea drinking behaviour, previously observed positive associations were often considered attributable to residual confounding from smoking.42,43 Our large sample size of never smokers provided us with adequate statistical power to investigate associations of interest among a population less likely to be influenced by smoking, although, as discussed above, passive smoking was not taken into consideration in the analysis of all cohorts. Our study is the first to investigate the association of lung cancer risk with combined exposure to coffee and tea. The generalizability and validity of our study is enhanced by involving diverse study populations with a wide spectrum of coffee and tea consumption and restricted to only prospectively collected data.
Our study has several limitations. First, coffee and tea intake were not uniformly measured across the cohorts, and type of coffee and tea information was not available in some cohorts. Second, data on coffee and tea brewing concentrations were not obtained, which may affect the amounts of bioactive compounds consumed. In addition, we were not able to account for potential changes in coffee and tea consumption and smoking habits after study enrolment in most cohorts. Also, participants’ coffee and tea consumption habits could have been different before entering the study and change afterwards. Those limitations may be overcome by applying genetic instruments for caffeine consumption in future research. Furthermore, there may be concerns that our results were heavily influenced by the NIH-AARP study because of its large sample size. Yet, the overall association of coffee drinking with lung cancer was generally similar across cohorts in our analysis. Furthermore, coffee drinking in the NIH-AARP was not associated with lung cancer among never smokers in the prior analysis,44 and this finding was confirmed when data from this cohort was analysed separately here, despite some analytic differences. Finally, although this is the largest study conducted to date, results from analyses in several subgroups, such as blacks or histologic subtypes in never smokers, were not reliable due to low statistical power. Future studies with larger sample sizes for those under-powered subgroup analyses are warranted to understand the associations to a more comprehensive extent.
In conclusion, this large-scale international pooling study found that higher consumption of coffee or tea was associated with an increased risk of lung cancer. We provided some evidence that the positive association between coffee consumption and lung cancer risk among current and former smokers could largely be explained by a residual confounding effect of smoking. However, residual confounding from smoking could not completely explain the positive associations of coffee and tea intake with lung cancer risk among lifetime never smokers, albeit weakly observed in this study. We caution that our findings should not be assumed to be causal due to imperfect adjustments for the effects of smoking and passive smoking. Additional studies are needed to fully understand the nature of the associations.
Supplementary Material
Novelty and Impact.
Most lung cancer cases included in previous studies investigating the associations of coffee and tea drinking with lung cancer risk were smokers and individuals of European descent. Never smokers and non-European populations were under-represented. This large-scale pooled analysis allowed an in-depth analysis on coffee and tea drinking, separately and jointly, by smoking status, race and cancer subtypes. Our study provides new information to this field and calls for further investigation on the nature of coffee- and tea-lung cancer associations.
Acknowledgement
The data used for this study were contributed by the NCI Cohort Consortium and Asian Cohort Consortium. The authors want to thank staff, investigators and participants of the contributing cohorts. We thank Dr. Mary Shannon Byers for her assistance in editing and preparing the manuscript.
Funding
This work was supported partially by a grant from the National Institutes of Health (R03 CA183021) and by the Ingram Cancer Professorship fund.
List of abbreviations:
- CI
Confidence Interval
- HPFS
Health Professionals Follow-Up Study
- HR
Hazard Ratio
- ICD
International Classification of Disease
- IWHS
Iowa Women’s Health Study
- NHS
Nurses’ Health Study
- NIH-AARP
National Institutes of Health-American Association of Retired Persons
- PLCO
Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial
- SCCS
Southern Community Cohort Study
- SMHS
Shanghai Men’s Health Study
- SWHS
Shanghai Women’s Health Study
- US
United States
Footnotes
Disclosure statement
The authors declare no conflict of interest.
Data accessibility
The data used in this study are available from the corresponding author upon reasonable request.
Ethics statement
All studies were approved by the Institutional Review Boards of their corresponding institutions, and informed consent was obtained by the parent studies.
Reference
- 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin 2019;69:7–34. [DOI] [PubMed] [Google Scholar]
- 2.Galarraga V, Boffetta P. Coffee Drinking and Risk of Lung Cancer-A Meta-Analysis. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cosponsored Am Soc Prev Oncol 2016;25:951–7. [DOI] [PubMed] [Google Scholar]
- 3.Wang L, Zhang X, Liu J, Shen L, Li Z. Tea consumption and lung cancer risk: a meta-analysis of case-control and cohort studies. Nutr Burbank Los Angel Cty Calif 2014;30:1122–7. [DOI] [PubMed] [Google Scholar]
- 4.Treur JL, Taylor AE, Ware JJ, McMahon G, Hottenga J-J, Baselmans BML, Willemsen G, Boomsma DI, Munafò MR, Vink JM. Associations between smoking and caffeine consumption in two European cohorts: Smoking and caffeine consumption. Addiction 2016;111:1059–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Swanson JA, Lee JW, Hopp JW. Caffeine and nicotine: a review of their joint use and possible interactive effects in tobacco withdrawal. Addict Behav 1994;19:229–56. [DOI] [PubMed] [Google Scholar]
- 6.Hang D, Kværner AS, Ma W, Hu Y, Tabung FK, Nan H, Hu Z, Shen H, Mucci LA, Chan AT, Giovannucci EL, Song M. Coffee consumption and plasma biomarkers of metabolic and inflammatory pathways in US health professionals. Am J Clin Nutr 2019;109:635–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zhao L-G, Li H-L, Sun J-W, Yang Y, Ma X, Shu X-O, Zheng W, Xiang Y-B. Green tea consumption and cause-specific mortality: Results from two prospective cohort studies in China. J Epidemiol 2017;27:36–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ishikawa A, Kuriyama S, Tsubono Y, Fukao A, Takahashi H, Tachiya H, Tsuji I. Smoking, alcohol drinking, green tea consumption and the risk of esophageal cancer in Japanese men. J Epidemiol 2006;16:185–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yang JJ, Yu D, Takata Y, Smith-Warner SA, Blot W, White E, Robien K, Park Y, Xiang Y-B, Sinha R, Lazovich D, Stampfer M, et al. Dietary Fat Intake and Lung Cancer Risk: A Pooled Analysis. J Clin Oncol Off J Am Soc Clin Oncol 2017;35:3055–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zheng W, Chow W-H, Yang G, Jin F, Rothman N, Blair A, Li H-L, Wen W, Ji B-T, Li Q, Shu X-O, Gao Y-T. The Shanghai Women’s Health Study: rationale, study design, and baseline characteristics. Am J Epidemiol 2005;162:1123–31. [DOI] [PubMed] [Google Scholar]
- 11.Shu X-O, Li H, Yang G, Gao J, Cai H, Takata Y, Zheng W, Xiang Y-B. Cohort Profile: The Shanghai Men’s Health Study. Int J Epidemiol 2015;44:810–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hankin JH, Stram DO, Arakawa K, Park S, Low SH, Lee HP, Yu MC. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001;39:187–95. [DOI] [PubMed] [Google Scholar]
- 13.Yoo KY, Shin HR, Chang SH, Lee KS, Park SK, Kang D, Lee DH. Korean Multi-center Cancer Cohort Study including a Biological Materials Bank (KMCC-I). Asian Pac J Cancer Prev APJCP 2002;3:85–92. [PubMed] [Google Scholar]
- 14.Nakaya N, Tsubono Y, Hosokawa T, Nishino Y, Ohkubo T, Hozawa A, Shibuya D, Fukudo S, Fukao A, Tsuji I, Hisamichi S. Personality and the risk of cancer. J Natl Cancer Inst 2003;95:799–805. [DOI] [PubMed] [Google Scholar]
- 15.Kim J Cancer screenee cohort study of the National Cancer Center in South Korea. Epidemiol Health 2014;36:e2014013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fukao A, Tsubono Y, Komatsu S, Tsuji I, Minami Y, Hisamichi S, Hosokawa T, Kawamura M, Takano A, Sugahara N, Ikeda T, Nishikori M. A Cohort Study on the Relation of Lifestyle, Personality and Biologic Markers to Cancer in Miyagi, Japan : Study Design, Response Rate and Profiles of the Cohort Subjects. J Epidemiol [Internet] 1995. [cited 2018 May 29];5:153–7. Available from: http://joi.jlc.jst.go.jp/JST.Journalarchive/jea1991/5.153?from=CrossRef [Google Scholar]
- 17.Sado J, Kitamura T, Kitamura Y, Zha L, Liu R, Sobue T, Nishino Y, Tanaka H, Nakayama T, Tsuji I, Ito H, Suzuki T, et al. Rationale, design, and profile of the Three-Prefecture Cohort in Japan: A 15-year follow-up. J Epidemiol 2017;27:193–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Li Q, Kakizaki M, Kuriyama S, Sone T, Yan H, Nakaya N, Mastuda-Ohmori K, Tsuji I. Green tea consumption and lung cancer risk: the Ohsaki study. Br J Cancer 2008;99:1179–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tsuji I, Nishino Y, Ohkubo T, Kuwahara A, Ogawa K, Watanabe Y, Tsubono Y, Bando T, Kanemura S, Izumi Y, Sasaki A, Fukao A, et al. A prospective cohort study on National Health Insurance beneficiaries in Ohsaki, Miyagi Prefecture, Japan: study design, profiles of the subjects and medical cost during the first year. J Epidemiol 1998;8:258–63. [DOI] [PubMed] [Google Scholar]
- 20.Tsugane S, Sawada N. The JPHC study: design and some findings on the typical Japanese diet. Jpn J Clin Oncol 2014;44:777–82. [DOI] [PubMed] [Google Scholar]
- 21.Hankin JH, Stram DO, Arakawa K, Park S, Low SH, Lee HP, Yu MC. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001;39:187–95. [DOI] [PubMed] [Google Scholar]
- 22.Sugiyama K, Sugawara Y, Tomata Y, Nishino Y, Fukao A, Tsuji I. The association between coffee consumption and bladder cancer incidence in a pooled analysis of the Miyagi Cohort Study and Ohsaki Cohort Study: Eur J Cancer Prev [Internet] 2017. [cited 2019 May 22];26:125–30. Available from: http://content.wkhealth.com/linkback/openurl?sid=WKPTLP:landingpage&an=00008469-201703000-00003 [DOI] [PubMed] [Google Scholar]
- 23.Tsubono Y, Nishino Y, Komatsu S, Hsieh C-C, Kanemura S, Tsuji I, Nakatsuka H, Fukao A, Satoh H, Hisamichi S. Green Tea and the Risk of Gastric Cancer in Japan. N Engl J Med [Internet] 2001. [cited 2019 May 22];344:632–6. Available from: 10.1056/NEJM200103013440903 [DOI] [PubMed] [Google Scholar]
- 24.Inoue M, Kurahashi N, Iwasaki M, Shimazu T, Tanaka Y, Mizokami M, Tsugane S, Japan Public Health Center-Based Prospective Study Group. Effect of coffee and green tea consumption on the risk of liver cancer: cohort analysis by hepatitis virus infection status. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cosponsored Am Soc Prev Oncol 2009;18:1746–53. [DOI] [PubMed] [Google Scholar]
- 25.Ahn Y, Kwon E, Shim JE, Park MK, Joo Y, Kimm K, Park C, Kim DH. Validation and reproducibility of food frequency questionnaire for Korean genome epidemiologic study. Eur J Clin Nutr 2007;61:1435–41. [DOI] [PubMed] [Google Scholar]
- 26.Yang P, Zhang X-Z, Zhang K, Tang Z. Associations between frequency of coffee consumption and osteoporosis in Chinese postmenopausal women. Int J Clin Exp Med 2015;8:15958–66. [PMC free article] [PubMed] [Google Scholar]
- 27.Yu Q, Liu Z-H, Lei T, Tang Z. Subjective evaluation of the frequency of coffee intake and relationship to osteoporosis in Chinese men. J Health Popul Nutr 2016;35:24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Smiciklas-Wright H, Mitchell DC, Mickle SJ, Goldman JD, Cook A. Foods commonly eaten in the United States, 1989–1991 and 1994–1996: Are portion sizes changing? J Am Diet Assoc 2003;103:41–7. [DOI] [PubMed] [Google Scholar]
- 29.Nechuta S, Shu X-O, Li H-L, Yang G, Ji B-T, Xiang Y-B, Cai H, Chow W-H, Gao Y-T, Zheng W. Prospective cohort study of tea consumption and risk of digestive system cancers: results from the Shanghai Women’s Health Study. Am J Clin Nutr 2012;96:1056–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.de Leon J, Diaz FJ, Rogers T, Browne D, Dinsmore L, Ghosheh OH, Dwoskin LP, Crooks PA. A pilot study of plasma caffeine concentrations in a US sample of smoker and nonsmoker volunteers. Prog Neuropsychopharmacol Biol Psychiatry 2003;27:165–71. [DOI] [PubMed] [Google Scholar]
- 32.Plowchalk DR, Rowland Yeo K. Prediction of drug clearance in a smoking population: modeling the impact of variable cigarette consumption on the induction of CYP1A2. Eur J Clin Pharmacol 2012;68:951–60. [DOI] [PubMed] [Google Scholar]
- 33.Bjørngaard JH, Nordestgaard AT, Taylor AE, Treur JL, Gabrielsen ME, Munafò MR, Nordestgaard BG, Åsvold BO, Romundstad P, Davey Smith G. Heavier smoking increases coffee consumption: findings from a Mendelian randomization analysis. Int J Epidemiol 2017;46:1958–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Treur JL, Taylor AE, Ware JJ, McMahon G, Hottenga J-J, Baselmans BML, Willemsen G, Boomsma DI, Munafò MR, Vink JM. Associations between smoking and caffeine consumption in two European cohorts. Addict Abingdon Engl 2016;111:1059–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Larose TL, Guida F, Fanidi A, Langhammer A, Kveem K, Stevens VL, Jacobs EJ, Smith-Warner SA, Giovannucci E, Albanes D, Weinstein SJ, Freedman ND, et al. Circulating cotinine concentrations and lung cancer risk in the Lung Cancer Cohort Consortium (LC3). Int J Epidemiol 2018;47:1760–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hamling JS, Coombs KJ, Lee PN. Misclassification of smoking habits: An updated review of the literature. World J Meta-Anal 2019;7:31–50. [Google Scholar]
- 37.Mojska H, Gielecińska I. Studies of acrylamide level in coffee and coffee substitutes: influence of raw material and manufacturing conditions. Rocz Panstw Zakl Hig 2013;64:173–81. [PubMed] [Google Scholar]
- 38.Manjanatha MG, Guo L-W, Shelton SD, Doerge DR. Acrylamide-induced carcinogenicity in mouse lung involves mutagenicity: cII gene mutations in the lung of big blue mice exposed to acrylamide and glycidamide for up to 4 weeks. Environ Mol Mutagen 2015;56:446–56. [DOI] [PubMed] [Google Scholar]
- 39.Cornelis MC, Kacprowski T, Menni C, Gustafsson S, Pivin E, Adamski J, Artati A, Eap CB, Ehret G, Friedrich N, Ganna A, Guessous I, et al. Genome-wide association study of caffeine metabolites provides new insights to caffeine metabolism and dietary caffeine-consumption behavior. Hum Mol Genet 2016;ddw334. [DOI] [PubMed] [Google Scholar]
- 40.Abd El-Aty AM, Choi J-H, Rahman MM, Kim S-W, Tosun A, Shim J-H. Residues and contaminants in tea and tea infusions: a review. Food Addit Contam Part Chem Anal Control Expo Risk Assess 2014;31:1794–804. [DOI] [PubMed] [Google Scholar]
- 41.Hossain MZ, Gilbert SF, Patel K, Ghosh S, Bhunia AK, Kern SE. Biological clues to potent DNA-damaging activities in food and flavoring. Food Chem Toxicol Int J Publ Br Ind Biol Res Assoc 2013;55:557–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Schwarz B, Bischof HP, Kunze M. Coffee, tea, and lifestyle. Prev Med 1994;23:377–84. [DOI] [PubMed] [Google Scholar]
- 43.Rodenburg EM, Eijgelsheim M, Geleijnse JM, Amin N, van Duijn CM, Hofman A, Uitterlinden AG, Stricker BH, Visser LE. CYP1A2 and coffee intake and the modifying effect of sex, age, and smoking. Am J Clin Nutr 2012;96:182–7. [DOI] [PubMed] [Google Scholar]
- 44.Guertin KA, Freedman ND, Loftfield E, Graubard BI, Caporaso NE, Sinha R. Coffee consumption and incidence of lung cancer in the NIH-AARP Diet and Health Study. Int J Epidemiol 2016;45:929–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.