|Home | About | Journals | Submit | Contact Us | Français|
Tobacco products are major contributors of exogenous N-nitroso compounds, a group of potent neurocarcinogens. Overall results from studies of smoking and brain tumors have been null, but have provided little information on duration, age at smoking initiation, and latency. We prospectively examined the relation between cigarette smoking and glioma risk among men and women in three large U.S. cohort studies: the Health Professionals Follow-up Study (HPFS), the Nurses’ Health Study I (NHS), and NHS II. Information on smoking history was obtained at baseline (1986 for HPFS, 1976 for NHS, and 1989 for NHS II) and updated biennially through 2002 in the HPFS and the NHS and 2003 in the NHS II. We confirmed 110 incident gliomas among men and 255 gliomas among women during 667,673 and 4,388,515 person-years of follow-up, respectively. Cox proportional hazard models were used to estimate incidence rate ratios and 95% confidence intervals between smoking and glioma risk adjusting for age, total meat intake, and alcohol and coffee consumption. Estimates from each cohort were pooled using a random-effects model after determining that there was no heterogeneity by sex. No association with glioma risk was observed between baseline or updated smoking status, intensity, duration, or age at smoking initiation among men and women. Furthermore, no association with glioma risk was observed after allowing for an induction period between smoking and glioma diagnosis. These findings provide strong evidence that cigarette smoking is not associated with an appreciably elevated risk of adult glioma.
Gliomas are the most common adult brain tumor and are associated with high fatality rates. Despite substantial efforts, the etiology of this cancer remains poorly understood. Presently, the only established risk factors for glioma are age, male sex, Caucasian race, and inherited factors, none of which are modifiable. Incidence rates are higher in industrialized nations, including the United States, Canada, Australia, and the United Kingdom, than in developing nations.1 Although access to diagnostic technology and medical expertise explains some of the geographic variation in incidence rates, it is likely that environmental factors also influence the development of gliomas.
Cigarette smoking is a plausible behavioral exposure that might modulate glioma risk. Tobacco products are major contributors of exogenous N-nitroso compounds; results from experimental studies have shown N-nitroso compounds to be potent neurocarcinogens2–4 that can induce glioma formation in rats.5
Both case-control6–21 and cohort22–25 studies have evaluated the relationship between cigarette smoking and the risk of adult glioma. In addition to smoking status, the majority of studies included information on dosage (cigarettes per day) or intensity (pack-years)6,8,11–14,16–21, 23–25; however, few included further information on duration of smoking,11,13,17,18,25 age at smoking initiation,13,17,18,25 years since quitting,11,18,25 and type of cigarette.8,11,18,19,24 Overall, no association has been observed between these cigarette-smoking exposures and glioma risk. In contrast, two recent cohort studies reported elevated risks of glioma among women with increasing numbers or packs of cigarettes smoked per day,24,25 increasing numbers of years smoked,25 and earlier age at smoking initiation25 and among former smokers who recently quit smoking.25 However, information on smoking was collected only at baseline.
Given the limited data on duration, age at smoking initiation, and latency in previous studies, and recent positive findings from two cohort studies, we evaluated the role of cigarette smoking and the risk of glioma by using three large prospective studies of men and women with detailed and updated smoking information, 365 glioma cases, and 26 years of follow-up.
The Nurses’ Health Study (NHS) was initiated in 1976, when 121,700 registered U.S. female nurses, aged 30–55 years, returned a mailed questionnaire that assessed information on lifestyle factors and medical and smoking histories.26,27 Similarly, the Health Professionals Follow-up Study (HPFS) is a cohort of 51,529 U.S. male physicians, dentists, optometrists, osteopaths, podiatrists, pharmacists, and veterinarians who were 40–75 years of age at enrollment in 1986.28 The study design and methods of dietary assessment and follow-up for the Nurses’ Health Study II (NHS II) are very similar to those for the NHS. In 1989, 116,686 women aged 25–42 years and living in 14 U.S. states were enrolled into the NHS II. Follow-up questionnaires are mailed biennially to all cohort members to update information on lifestyle factors and newly diagnosed medical conditions. The follow-up rate for the cohorts for incidence of cancer was greater than 95% of the total possible person-years. The questionnaire response rate over the period of follow-up in the cohorts is approximately 90%.
Information on smoking was obtained at baseline (1976 for NHS, 1986 for HPFS, and 1989 for NHS II): Past smokers reported when they last smoked, and current smokers reported intensity of smoking (cigarettes per day). In addition, at baseline in the HPFS and the NHS II, we inquired about the average number of cigarettes smoked per day before age 15 years and at ages 15–19, 20–29, 30–39, 40–49, 50–59, and 60 and older. In the NHS, we inquired at what age a smoker had begun smoking and how many cigarettes per day were smoked during the first five years of smoking. For past smokers, we asked the ages of initiation and quitting, and the number of cigarettes per day while smoking. We assumed that a past smoker had smoked the same level throughout the reported years of cigarette use. Past smokers are individuals who at baseline indicated having smoked more than 20 or more packs of cigarettes in their lifetime or who quit smoking during follow-up. Current smoking status and intensity of smoking were updated biennially in subsequent questionnaires for all cohort members. We estimated pack-years (the equivalent of smoking 20 cigarettes a day for one year) by multiplying the number of packs smoked per day by the number of years of smoking (<10, 10–24, 25–44, and 45 or more pack-years). The time since quitting was calculated for those who quit during follow-up (<10 and 10 years). Categories were derived for age at start of smoking (19 and 20 years), based on the earliest decade where smoking was recorded. Pipe or cigar smoking was assessed among men only at baseline in 1986. Use of a filter cigarette was assessed at baseline in 1978 among women in the NHS cohort only.
Food-frequency questionnaires were initially collected in 1980 for the NHS, 1986 for the HPFS, and 1991 for the NHS II, and diet was generally updated every four years. The food-frequency questionnaires have previously been shown to validly assess dietary and alcohol intake during the past year.29–34
On each biennial questionnaire, participants were asked whether they had been diagnosed with any cancer, heart disease, or other medical conditions during the previous two years. When permission was received from the subjects (or next of kin for decedents), medical records (including pathology reports) were obtained from hospitals and reviewed by study investigators blinded to questionnaire exposure information. Nonrespondents were telephoned in an attempt to confirm the initial cancer report and date of diagnosis. Medical records are requested for reported and deceased glioma patients; 88% of glioma diagnoses were confirmed by medical or death records (<3%). When we were unable to obtain medical records, we attempted to corroborate diagnoses of glioma with additional information from the participant or next of kin or by cross-linking with cancer registries. We included only cases for which a medical or death record or other confirmation of the cancer was obtained. We included all glioma brain tumors: astrocytoma, glioblastoma, oligodendroglioma, ependymoma, and mixed-glioma subtypes. Vital status was ascertained through next of kin and the National Death Index; these methods identify at least 98% of deaths in the cohorts.35
We confirmed 110 newly diagnosed gliomas between 1986 and 2002 among men, 230 gliomas among women in the NHS between 1976 and 2002, and 25 gliomas among women in the NHS II between 1989 and 2003.
Person-time of follow-up was calculated from the date for return of the baseline questionnaire (1986 for HPFS, 1976 for NHS, and 1989 for NHS II) until the date of glioma diagnosis, date of death from any cause, or the end of follow-up (December 31, 2002, for HPFS; May 31, 2002, for NHS; and May 31, 2003, for NHS II), whichever came first. After excluding participants who reported a history of cancer other than nonmelanoma skin cancer or those with missing information on smoking at baseline, the cohorts for analyses included 46,327 men in the HPFS whose cases were followed for up to 16 years (667,673 person-years of follow-up), 115,087 women whose cases were followed for up to 26 years in the NHS, and 96,504 women whose cases were followed for up to 14 years in the NHS II (4,388,515 total person-years of follow-up among women).
Cox proportional hazards models for failure-time data were used to estimate the incidence rate ratio (RR) and 95% confidence interval (CI) for glioma risk and to adjust simultaneously for age (one year) and other potential risk factors, including total meat intake (consisting of processed meats; bacon; hot dogs; hamburger; beef, pork, or lamb as a sandwich or mixed dish; beef, pork, or lamb as a main dish; chicken with skin; and chicken without skin [quintiles]), alcohol consumption (0, 0.1–1.4, 1.5–4.9, 5.0–29.9, and 30.0 g/day), and coffee consumption (0, 1, 2–3, and 4 cups per day). Additional adjustment for potential risk factors, including processed meat intake (consisting of processed meats, bacon, and hot dogs [quintiles]), total intake of vitamins C or E (milligrams per day; energy-adjusted vitamin intake from diet and vitamin supplement; continuous), multivitamin supplement use (yes/no), state of residence in the United States (West, Midwest, South, and Northeast), body mass index (18.0–22.9, 23.0–24.9, 25.0–26.9, 27.0–29.9, and 30.0 kg/m2), height (inches [quintiles]); type of profession (among men only; pharmacist, specialist [optometrist or podiatrist], physician, veterinarian, or dentist), and reproductive factors (status and age at menopause [among women who had a natural menopause or among women with bilateral oopherectomy, age at surgery]: premenopausal; post-menopausal, age < 45; postmenopausal, age 45–49; postmenopausal, age 50–55; and postmenopausal, age > 55) did not change the associations of cigarette smoking with glioma risk. Because of the relative homogeneity of the population of male health professionals and female nurses, it was unnecessary to control for education or socioeconomic status. All covariates were assessed at baseline and repeatedly assessed in subsequent questionnaires for all cohort members (similar to smoking information), and updated analyses used time-varying covariates in the Cox proportional hazards models. For example, in the HPFS, covariate data from the 1986 questionnaire were used for follow-up from 1986 to 1990, data from the 1990 questionnaire were used for follow-up from 1990 to 1994, data from the 1994 questionnaire were used for follow-up from 1994 to 1998, and data from the 1998 questionnaire were used for follow-up from 1998 to 2002. Details of this method are described elsewhere.36,37
Additional analyses were restricted by tumor histology (astrocytoma [ICD-O: 94003, 94013, 94113, 94103, 94203, and 94213] or glioblastoma [ICD-O: 94403, 94413, and 94423]). We also conducted analyses designed to estimate the risk of glioma after different time periods after the onset of smoking (i.e., the induction period). Tests for (multiplicative) interaction by age or consumption of alcohol were performed by examining stratum-specific estimates and formally by use of likelihood-ratio tests. Tests of linear trend were conducted by using the continuous variable for pack-years of smoking, using Cox proportional hazards regression.
Before pooling RR and trend test probability (p) values by using meta-analysis, tests of heterogeneity of the main exposures by cohort were performed using the Q statistic, and data among men in the HPFS and women in the NHS and the NHS II were pooled using a random- effects model for the log of the RR;38 no statistically significant heterogeneity was observed (p values for heterogeneity for each category were > 0.9). All reported p values are two-tailed.
The mean (±SD) age of glioma patients was 65.3 (10.0) years for men and 59.3 (9.2) years and 39.9 (5.6) years for women in the NHS and the NHS II, respectively. The crude incidence rate of glioma was 16 per 100,000 person- years among men in the HPFS, 9 per 100,000 person-years among women in the NHS, and 1 per 100,000 person-years among women in the NHS II (similar to CBTRUS data39). At baseline, as expected, current smokers were more likely to drink alcohol or coffee and were less likely to consume fruits and vegetables (Table 1). For both men and women, their intake of total meat, their height, and their body mass index did not vary appreciably across categories of smoking history.
Overall, we observed no association between cigarette smoking and the risk of glioma among men and women (Table 2). Compared to never smokers, no association with glioma risk was observed for smoking status, including time since quit for past smokers, pack-years of smoking, duration of smoking, or age at initiating smoking, after adjusting for age, total meat intake, and alcohol and coffee consumption among men and women. Furthermore, no association with glioma risk was observed for daily number of cigarettes smoked for current smokers among men and women (pooled multivariable RR [95% CI] = 1.49 (0.93–2.38) for fewer than 15 cigarettes per day and 1.12 [0.74–1.69] for 15 or more cigarettes per day). Risk estimates did not change appreciably after additional inclusion of other potential risk factors into the statistical models. Similar results were observed when using baseline cigarette smoking (1986 for HPFS, 1976 for NHS, and 1989 for NHS II; data not shown).
We also examined whether cigarette smoking at earlier or later ages was related to glioma risk. Overall, no association was observed between total pack-years of smoking before or after the age of 30 with the risk of glioma among men and women (Table 3).
To estimate the induction period between smoking and glioma diagnosis, we evaluated the association between the time since onset of smoking and the risk of glioma among men and women; overall, the risk of glioma appeared unrelated to smoking up to at least 40 or more years after initiation of smoking (pooled p value for trend = 0.29). Furthermore, risk estimates for glioma according to time since starting smoking did not vary by dose (daily number of cigarettes at the start; data not shown).
We collected information on glioma histology or location for 82 men (75%) and 202 women (79%). In the cohorts, approximately 95% and 88% of brain tumors among men and women, respectively, were classified as astrocytoma; the remaining brain tumors were classified as oligodendroglioma, ependymoma, or mixed-glioma subtypes. For both men (82%) and women (65%), glioblastoma was the most common histologic type. Glioma was most commonly found in the frontal lobe (25%) and temporal lobe (20%). Additional analyses conducted among more homogeneous cancer subgroups showed similar null results between updated cigarette smoking and the risk of astrocytoma or glioblastoma. Furthermore, no relation between cigarette smoking and glioma risk was observed by anatomical site (frontal lobe [73 cases] or temporal lobe [53 cases], separately) or among cases confirmed by pathology reports (225 cases; data not shown).
The association between updated cigarette smoking and glioma risk was examined across strata of age and alcohol consumption; there was no evidence that the association was modified by any of the exposures evaluated (pooled p values for interaction were 0.57 and 0.93 for age and alcohol, respectively).
We could not properly evaluate the use of a nonfiltered cigarette (among women in the NHS only; assessed in 1978) or updated pipe or cigar smoking (among men only) because of the small number of glioma cases among users of nonfiltered cigarettes (five cases) or pipe or cigar smokers (one case).
This study represents the first prospective study to examine the risk of adult glioma by using periodically updated and detailed information on cigarette smoking in three well-established cohorts of men and women with a large number of glioma cases. Overall, we find no association between cigarette smoking and the risk of adult glioma. No associations with glioma risk were observed between baseline or updated smoking status, dosage, duration, or age at smoking initiation among men and women. Moreover, we observed no association after allowing for a range of possible induction periods between smoking and glioma diagnosis, or after restricting analyses among more homogeneous cancer subtypes. In addition, there was no evidence that the association was modified by age or alcohol consumption.
Most studies suggest no overall association between smoking status or dosage of cigarette smoking and the risk of adult brain tumors10,21,23 or gliomas.6–9,11–20,22,24,25 In contrast, a retrospective cohort study that included 130 gliomas24 observed an increased risk of glioma among former or current smokers of more than two packs of cigarettes per day compared to never smokers (RR = 2.3; 95% CI, 1.2–4.5); however, the association was limited to women only (p value for trend = 0.04). Similarly, a cohort study of women recruited into the Canadian National Breast Screening Study that included 117 gliomas25 observed a suggestive dose-response association between glioma risk and number of cigarettes smoked per day (p value for trend = 0.08) or pack-years of cigarette smoking (p value for trend = 0.07), compared to never smokers.
One case-control study observed an increased glioma risk among men who had smoked for less then 10 years (250 gliomas among men; RR = 2.49; 95% CI, 1.25–4.92),17 and a cohort study observed a suggestive dose-response association between glioma risk and number of years smoked (p value for trend = 0.06)25 compared to never smokers; however, null associations have also been observed between glioma risk and duration of smoking in three case-control studies.11,13,18
Similarly mixed results have been observed for age at smoking initiation and the risk of glioma. Early age at smoking initiation was associated with an elevated risk of glioma in one study among women (RR = 1.67; 95% CI, 1.03–2.72, for age < 20 compared to never smokers),25 but the opposite was reported for men in another study (RR = 2.72; 95% CI, 1.48–5.02, for age 20 compared to never smokers), but not among women.17 Two additional studies with data on age at smoking initiation reported no association.13,18
Null associations between glioma risk and years since quitting, as we found, have also been observed in other studies;11,18 however, one prospective cohort study reported that women who quit smoking more than 10 years prior to baseline were at a decreased risk of glioma compared to those who had quit within the 10 years prior to baseline (RR = 0.39; 95% CI, 0.19–0.82).25
Overall, no association has been observed between type of cigarette smoked (i.e., filtered, unfiltered, regular, king-size, long, mentholated, and plain) and glioma risk,11,18,19,24 with one exception.8 One study evaluated possible modification of cigarette smoking status (packs of cigarettes per day) and glioma risk by alcohol use.24 Glioma risk among current nondrinkers of alcohol was elevated across all smoking levels; however, this interaction was not statistically significant.
The difficulty in comparing results across studies of smoking and the risk of glioma may be due to several limitations. Most epidemiological studies have relied on case-control studies of brain tumors and are particularly prone to methodological biases, including small number (<100) of glioma cases,6,7,18,21 potential recall or selection bias,6–11,13,16,20,21 and measurement error due to surrogate sources of information.7,8,11,12,14–17,19 Furthermore, the potential misclassification of smoking habits over time in prospective studies because of the use of baseline information only22–25 and, regardless of study design, the lack of studies with detailed smoking exposures (i.e., age at first start, duration, time since quit, and latency) make interpretation of findings across studies difficult.
The strengths of our study include its large sample size, the prospective design, long follow-up, and analyses based on confirmed cases. Compared to previous prospective studies of cigarette smoking and the risk of incident glioma,22,24,25 the current study is the first to evaluate the risk of glioma prospectively with detailed periodically updated information on cigarette smoking over the period of follow-up, which may reduce misclassification by accounting for changes in smoking habits over time. In addition, only one other cohort study25 has evaluated the association between cigarette smoking status, dosage, duration, age at start, and glioma risk, and no other study has examined latency of smoking. The prospective design precludes recall bias, and potential selection bias is minimized by the very high rate of follow-up over a long period. No proxies were needed because information on smoking was obtained before the occurrence of disease. Participants are not a random sample of U.S. men and women, so the findings might not be directly generalizable to the U.S. population as a whole; however, it is unlikely that the relationship with gliomas among participants in these cohorts will differ from men and women in general. Because of the small number of cases among users of nonfiltered cigarettes or pipe or cigar smokers, we could not evaluate these associations with glioma risk.
In conclusion, our findings provide strong evidence that cigarette smoking is not associated with risk of adult glioma.
This work has been supported by grants CA98566, CA110948, CA55075, and CA87969 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. We thank Dr. Walter Willett for his valuable advice and Barbara Vericker and Barbara Egan for their assistance.