Annals of Oncology Advance Access originally published online on May 25, 2006
Annals of Oncology 2006 17(8):1283-1289; doi:10.1093/annonc/mdl096
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2006 European Society for Medical Oncology
Cigarette smoking and alcohol consumption as determinants of survival in non-Hodgkin's lymphoma: a population-based study
1 University of Milan and Department of Medicine and Medical Specialties, IRCCS Maggiore Hospital, Mangiagalli and Regina Elena Foundation, Milan, Italy; 2 Occupational and Environmental Epidemiology Unit, Center for Study and Prevention of Cancer, Florence, Italy; 3 Epidemiology Unit, National Cancer Institute, Milan, Italy; 4 Unit of Biostatistics and Clinical Trials, Romagnolo Oncogic Institute, Forlì, Italy; 5 Environmental Epidemiology and Biostatistics Unit, National Cancer Research Institute, Genoa, Italy; 6 Cancer Registry, Ragusa Hospital, Ragusa, Italy; 7 Cancer Epidemiology Unit, University of Turin, Turin, Italy; 8 Department of Epidemiology and Public Health, Imperial College, London, UK
* Correspondence to: Dr. T. Battaglioli, Department of Medicine and Medical Specialties, University of Milan and IRCCS Maggiore Hospital, Mangiagalli and Regina Elena Foundation, Via Pace 9, 20123 – Milan, Italy. Tel: +39 0255035468; Fax: +39 02 50320723; E-mail: tullia.battaglioli{at}unimi.it
 |
Abstract |
|---|
 Top Abstract introductionmaterials and methodsresultsdiscussionReferences |
|---|
Background: The risk of non-Hodgkin's lymphoma (NHL) seems to be enhanced by cigarette smoking and lowered by alcohol drinking.
Patients and methods: To assess whether cigarette smoking and alcohol drinking affect NHL survival, a population-based prospective study on 1138 Italian patients, diagnosed in 1991–1993, followed-up until 2002, was carried out. At diagnosis, clinical and socio-demographic data were recorded and lifestyle habits were assessed through a validated questionnaire. Survival analysis was performed with Kaplan-Meier methods. Hazard ratios (HR) were estimated by Cox regression.
Results: The mean follow-up was 6.6 years (standard deviation (SD) 4.3). The mean survival time was 7.56 years (SD 0.155). At both univariate and multivariate analysis heavy cigarette smoking and alcohol drinking were associated with poor survival. Compared with those with a lower cumulative exposure to tobacco smoking, those who had smoked >31 pack-years had a worse survival (HR = 1.60, 95%CI = 1.18–2.18). Drinkers had a higher risk of death compared with non-drinkers (HR = 1.41, 95%CI = 1.10–1.81). Considering only those who had NHL as cause of death, the HR for the higher category of pack-years smoked, compared with the lowest, was 1.63 (95% CI = 1.15–2.33) and for drinkers, compared with non-drinkers, it was 1.33 (95% CI = 1.01–1.80).
Conclusions: cigarette smoking and alcohol drinking may influence NHL survival.
Key words: alcohol, non-Hodgkin's lymphoma, survival, smoking
|
introduction |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionReferences |
|---|
Recently, a pooled analysis on the relation between cigarette smoking and non-Hodgkin's lymphoma (NHL), reported that smoking may increase the risk of developing follicular lymphoma, but it does not seem to affect the risk of other NHL subtypes [1
]. On the contrary, some studies not included in the analysis, reported no association [2–5]. A pooled analysis on the link between alcohol consumption and NHL [6], reported that alcohol drinkers had a lower risk of NHL, in particular of Burkitt's, follicular, and diffuse lymphomas, whereas other studies [3, 7–11] detected no association. To date, there are no studies investigating the effect of cigarette smoking and alcohol consumption on NHL survival. Since they have been demonstrated to have a potential role in the causation of NHL, it is reasonable to hypothesize an effect on disease progression and thus survival. Because they seem to be mild risk factors, the availability of a large sample size is crucial to detect an effect. With the aim of investigating this still unexplored field, a population-based prospective study on a large number of individuals, followed-up over a substantial time period, was carried out.
|
materials and methods |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionReferences |
|---|
Patients were a subset of a large Italian case-control study in which all incident cases of NHL occurring in 1991–1993, in individuals aged 20–74 years, residents of 11 Italian areas (city of Turin, provinces of Verona, Novara, Vercelli, Varese, Imperia, Forlì, Florence, Siena, Latina, Ragusa), covering about 7 million people, were identified as follows: (i) systematic search in departments of hematology, general medicine, surgery and pathology in all hospitals within the study areas; (ii) systematic search in cancer registries covering the study areas; (iii) regular survey of nationally relevant medical centers (in Milan, Pavia, Rome and Bologna), to identify cases from the study areas. Detailed methods were previously reported [12
, 13]. This study refers to 1158 cases incident in six of the areas in which the case-control study was conducted. The general characteristics of the included areas are not different from the areas not included in the study.
All diagnoses were reviewed and confirmed by a single pathologist. NHLs were classified according to the Working Formulation [14]. For subtype analysis, WF subtypes were grouped as follows: Small Lymphocytic Lymphoma/Chronic Lymphocytic Leukaemia (SLL/CLL – WF category A), Follicular (WF categories B-D), Diffuse Large B Cell Lymphoma (DLBCL – WF categories G and H), and ‘Other’ (WF categories E, F, I, J and miscellaneous).
Gender, date of birth, date of diagnosis, study area and educational level (as a proxy for socio-economic status [SES]) were recorded. As soon as a suspected case was identified, a person-to-person interview by trained blinded personnel was conducted. When direct interview was not possible because of death or illness, proxy responders were interviewed. Information about tobacco smoking and ethanol consumption was retrieved with a standardized questionnaire. Each subject gave his/her informed consent to participate in the study. The participation rate was 82%. Patients were followed up until December 31, 2002. Information on vital status and cause of death was obtained through death certificates.
Individuals who had ever smoked >100 cigarettes or had smoked regularly for 
6 months were classified as ever smokers and further classified into current/former smokers according to their habits 6 months before NHL diagnosis. Cumulative lifetime exposure to smoking was computed as number of pack-years smoked, obtained by multiplying the number of daily packs of cigarettes by the years of smoking. The daily number of cigarettes and the number of smoking years were split respectively into three groups (cut-points at the 33rd and 66th percentiles of the distribution among smokers), and four categories (cut-points at the 16.5th, 33rd and 66th percentiles of the distribution among smokers). The pack-years smoked and the number of years elapsed since quitting smoking were split into 3 groups, with cut-points at the 33rd and 66th percentiles of the distribution among smokers.
Individuals who consumed >0.01 g of ethanol a month were considered drinkers. The average daily number of drinks up to 6 months before diagnosis was recorded. The ethanol content of each drink was estimated as follows: 1 can of beer/1 glass of fortified wine = 15 g of ethanol, 1 glass of wine = 13 g of ethanol, 1 glass of spirits = 10 g of ethanol. The average daily ethanol intake, the duration of consumption, and the cumulative lifetime alcohol consumption were split respectively into four categories (cut-points at 0 g [non-drinkers], and at the 33rd and 66th percentiles of the distribution among drinkers), into five categories (cut-points at 0 years, and at the 16.5th, 33rd and 66th percentiles of the distribution among drinkers), and into five groups (cut-points at 0 kg, and at the 25th, 50th and 75th percentiles of the distribution among drinkers). Information on duration of alcohol consumption was lacking for 87 drinkers.
Univariate survival analysis was performed by Kaplan-Meier methods. Differences among subgroups were investigated by the Log-rank test or, when needed, the Wilcoxon test [15]. Multivariate analysis was performed by Cox regression models. Gender, age, education and type of interview were included as a priori confounders in all models. Alcohol drinking was included a priori in all models for smoking, and vice versa. Any other variable and interaction terms, were added in the model if it significantly improved its fit (P > 0.1) or if it changed by at least 20% at least one of the coefficients. The assumption of proportionality of hazards (PH) was assessed plotting log-cumulative hazards of the subgroups of the covariate under investigation versus log-time. When the PH assumption did not hold for a covariate, separate baseline hazards were fitted for that covariate. Alternatively, follow-up was split in two (first 5 years and remaining follow-up) and if the PH assumption appeared to hold for the first 5 years of follow-up, only this first period was considered. Analyses for linear trend were done including the categories as continuous variables in Cox models.
All statistical tests were two-sided with an 
level of 0.05 and were performed using the statistical software S-Plus 6.0 (1988–2001 Insightful Corp, Lucent Technology Inc).
|
results |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionReferences |
|---|
Out of 1158 incident cases, 20 were excluded because of errors or relevant missing data in the medical records or in the questionnaire. Men were significantly younger than women. A direct interview was possible for 901 cases (79%). The years of smoking, the average daily number of cigarettes and the pack-years smoked were significantly higher in men (Table 1).
|
Seventy-nine percent of individuals were drinkers and 77% (693 cases) of them obtained at least 75% of the daily intake of ethanol from wine. Women were more likely to be non-drinkers and had a significantly lower mean daily intake of ethanol but a longer mean duration of drinking (Table 1).
During follow-up (mean = 6.6 years, standard deviation [SD] = 4.3), 641 individuals died: 77% (489 cases) from NHL, 23% (145 cases) from other causes. For 6 the cause of death was unknown. The overall mean survival time was 7.56 years (SD 0.155), median 8.04 years (95% confidence interval [95% CI] = 7.14–9.37). At univariate analysis, men had a shorter mean survival (7.03 years, SD = 0.206 versus 8.21 years, SD = 0.230, p < 0.001), and an older age was associated with a shorter mean survival (8.97 years, SD = 0.411 for the youngest versus 6.56 years, SD = 0.179 for the oldest, p < 0.001). Follicular NHL patients had the longest mean survival (8.69 years, SD = 0.434).
Table 2 shows survival analysis according to lifestyle habits. Ever and never smokers had a similar survival time. At multivariate analysis, the average daily number of cigarettes influenced survival only in women: compared with light smokers, a higher risk of death was found for moderate (Hazard Ratio [HR] = 2.15, 95% CI = 1.40–3.30) and heavy smokers (HR = 1.93, 95% CI = 1.29–2.90). The longer the duration of smoking, the higher was the risk of death. The number of pack-years had a clear influence on survival in both genders. Compared with individuals with the lowest cumulative exposure to smoke, those with a moderate exposure had a non-statistically significant higher risk of death (HR = 1.25, 95%CI = 0.93–1.68) and those with a heavy exposure had a statistically significant higher risk of death (HR = 1.60, 95%CI = 1.18–2.18), with a P for trend of 0.002. Among former smokers, compared with individuals who had quitted smoking 
7 years before diagnosis, those who had quitted >7 years before, had a non-statistically significant lower risk of death.
|
Drinkers had a shorter mean survival (7.35 years, SD = 0.174 versus 8.29 years, SD = 0.329, P = 0.008) and a higher risk of death at multivariate analysis (HR = 1.41, 95% CI = 1.10–1.81). A higher daily intake of ethanol was associated with a higher risk of death (P = 0.02). Among individuals who had drunk for more than 10 years, the longer duration of drinking the greater was the risk of death, but no statistically significant trend was found (P = 0.07). The higher the cumulative intake of ethanol, the higher was the risk of death (P = 0.02).
Across subtypes (Table 3), a higher average daily number of cigarettes and a higher number of pack-years smoked were associated with a higher risk of death, in particular for follicular lymphoma. In all subgroups, drinkers had a worse survival compared with non-drinkers. In follicular NHL, light drinkers had a HR (95% CI) of 1.98 (0.90–4.5), but the HR progressively decreased for higher intakes; the p for trend though, was not significant (P = 0.9).
|
Because smokers and drinkers may have a reduced survival due to reasons other than NHL, such as other neoplastic or cardiovascular diseases, the analysis was repeated considering only survivors and those who had NHL as cause of death: the effect of lifestyle habits did not substantially change. In particular, the HR for the higher category of pack-years smoked, was 1.63 (95% CI = 1.15–2.33) and the HR for drinkers was 1.33 (95% CI = 1.01–1.80).
|
discussion |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionReferences |
|---|
The observed survival of our study population was comparable with other population-based studies [16
, 17]. The results of this study suggest that cigarette smoking may influence survival of patients with NHL. In fact, individuals who had smoked more than 31 pack-years, had a 60% higher risk of death, compared with smokers with a nearly 50% lower cumulative exposure (14 pack-years). A significant trend was observed from the lowest to the highest exposure category, suggesting a dose–response relationship. There was also some evidence that the risk of death among former smokers may decrease with time since quitting smoking. The investigation of the effect of smoking in the histological subtypes, although hampered by smaller numbers, showed, in all subgroups, an association of a higher amount of cigarettes smoked and a higher cumulative exposure to cigarette smoking with a non-significantly higher risk of death, in particular for follicular lymphoma. Cigarette smoke contains several mutagenic and carcinogenic compounds and has been linked to the development of several tumors [18, 19]. An association of smoking with a higher prevalence of the bcl-2 oncogene translocation t(14;18), which leads to an increase of bcl-2 protein, which inhibits apoptosis, has been reported [20]. This mutation is highly prevalent in NHL patients, in particular of follicular subtype and a recent case-control study, although limited by small numbers, showed a non-significant higher risk of t(14:18) positive NHL among smokers, who in contrast did not have a higher risk of t(14:18) negative NHL [3]. Over-expression of bcl-2 is a negative prognostic factor in NHL, particularly DLBCL [21]. Smoking also contains aromatic hydrocarbons, which induce mutations in the p53 gene (a nuclear protein crucial for the regulation of cellular cycle and apoptosis), which is a potential prognostic factor in NHL [22]: patients with wild type p53 respond better to treatment and have better survival rates than patients with mutated p53 genes. To date, there are no studies investigating the possible link between smoking, molecular aberrations and NHL prognosis. Cigarette smoke also induces immunosuppression, in particular of T-cells [23, 24], which play an important role in the detection and elimination of malignant cells. It is also associated with a reduction of immunoglobulins [25] and natural-killer cells. The extent of changes in immune function from cigarette smoking is related to its duration and intensity [26]. The immunosuppression determined by smoking may enhance the pre-existing impairment of the immunological function of NHL patients, contributing to the creation of a microenvironment that promotes tumour growth and protects it from clearance. Smoking can influence survival from NHL not only through direct action on the disease, but also through more general detrimental effects on health. In fact, being more prone to infections, respiratory and cardiovascular diseases, smokers are also more frail patients, and may be more susceptible to the complications of chemotherapeutic regimens and less likely to be administered more aggressive – and so more efficient – therapies because of concomitant diseases. Unfortunately, detailed information on medical history and treatment of the participants was not available for this analysis. Overall, drinkers had a 41% increased risk of death compared with non-drinkers. An association between survival and daily ethanol intake, duration of consumption and cumulative intake was found, with heavier drinkers having a higher risk of death than non-drinkers but a lower risk compared to moderate drinkers. When the analyses were performed on the different subtypes, estimates became unstable because of smaller numbers. Nevertheless, SLL/CLL and DLBCL showed the same tendency as the overall study population. Considering follicular lymphoma, drinkers had a non-significant 57% higher risk of death than non-drinkers, but those with a higher ethanol intake or duration of consumption had a lower risk of death, compared with light drinkers. All the estimates though were not statistically significant.
Alcohol is oxidized to acetaldehyde, which is genotoxic [27] and appears to induce chromosomal aberrations in lymphocytes of alcoholics [27]. It also has immunomodulatory effects, which may be different according to the degree of intake: heavy drinkers may have an impaired immune function, but light/moderate drinkers may have an improved immune response [28]. Our study, which is the first investigating the effect of alcohol on NHL survival, did not detect a protective effect but a detrimental one. Not only heavy drinkers but also moderate and light drinkers were at higher risk, when compared with non-drinkers. The presence of an association with average daily intake of ethanol, duration of consumption, and cumulative intake suggests that the risk of death may be influenced not only by recent exposure but also by the overall consumption of alcohol. Unfortunately, this study did not have enough power to draw conclusions on the different histological subtypes. However, it suggested a higher risk for follicular lymphoma.
Drinking habits were recorded up to 6 months before diagnosis, in order to avoid the inclusion among non-drinkers of individuals who might have stopped drinking because of the disease. Nevertheless, some individuals classified as non-drinkers, may have been former drinkers. If some of them were ‘sick quitters’, who gave up drinking because of some disease which may have also influenced subsequent survival, the effect of alcohol would have been underestimated. Generally, self-report of smoking habits appears reliable [29]. Drinkers tend to underreport the amount of alcohol drunk, but this would have underestimated the exposure. Proxy responders may have been less likely than patients to recall details about smoking and drinking history, in particular in the distant past. However, the proportion of smokers and drinkers among directly and indirectly interviewed individuals was the same. If proxy responders recalled less precisely the lifestyle habits of their next of kin, their effect would have been underestimated. In order to minimize the possibility of exposure misclassification, interviewers were trained through a dedicated course and a standardized structured questionnaire was used. Interviews were performed as soon as possible, even before confirmation of diagnosis in suspected cases, and this should also have reduced the possibility of overestimation of cigarette or alcohol intake linked to awareness of the nature of the disease by patients. We did not collect information on stage at diagnosis and/or therapies, but it is unlikely that such variables confound the association with smoking and alcohol drinking. In fact, if such confounding effect exists, it should be indirect, mediated by social class (i.e. low SES is associated with both tobacco smoking/alcohol drinking and stage at diagnosis), but we have carefully controlled for SES. The possibility of adjustment for confounders such as drinking habits for smoking (and vice versa) or SES, is another strength of this study. Selection bias is very unlikely because the study was population-based with a very high response rate (82%). A limit is that patients were interviewed only at entrance in the study, and some subjects may have changed their habits during follow-up. Nevertheless, if some change has occurred, it is more likely that it was towards giving up or reducing smoking and/or drinking because of sickness, leading to underestimation of exposure. All diagnoses were centrally reviewed by a single experienced pathologist, who classified them according to the WF. A random sample of 25% of diagnoses was validated by a panel of pathologists. Because subtype analysis was done collapsing WF subgroups, it is possible that some degree of disease misclassification has occurred. To minimize this we used only the histological subtypes of the WF for which a satisfying reliability was shown by studies addressing this issue by means of expert pathological review [30], grouping the others in the ‘other’ category.
In conclusion, this study shows that heavy cigarette smoking and ethanol drinking may negatively affect survival from NHL. The analysis of the effect on the different histological subtypes suggests a different effect on survival of alcohol drinking in some histological subtypes. Future research should focus on the attempt to gather a higher number of patients for each histological subtype of NHL.
|
Acknowledgements |
|---|
Many thanks to Professor Nicky Best, Department of Epidemiology and Public Health, Imperial College, London, UK, for the valuable statistical advice and support and to Simonetta di Lollo, University of Florence, Italy, for her important help.
Received for publication January 29, 2006. Revision received March 17, 2006. Accepted for publication March 28, 2006.
|
References |
|---|
|
TopAbstractintroductionmaterials and methodsresultsdiscussionReferences |
|---|
1. Morton LM, Hartge P, Holford TR et al. Cigarette smoking and risk of non-Hodgkin lymphoma: a pooled analysis from the International Lymphoma Epidemiology Consortium (InterLymph). Cancer Epidemiol Biomarkers Prev 2005; 14: 925–933.
2. Adami J, Nyren O, Bergstrom R et al. Smoking and the risk of leukemia, lymphoma, and multiple myeloma (Sweden). Cancer Causes Control 1998; 9: 49–56.[CrossRef][ISI][Medline]
3. Schroeder JC, Olshan AF, Baric R et al. A case-control study of tobacco use and other non-occupational risk factors for T(14; 18) subtypes of non-Hodgkin's lymphoma (United States). Cancer Causes Control 2002; 2: 159–168.
4. Chiu BC-H, Weisenburger DD, Cantor KP et al. Alcohol consumption, family history of hematolymphoproliferative cancer, and the risk of non-Hodgkin's lymphoma in men. Ann Epidemiol 2002; 12: 309–315.[CrossRef][ISI][Medline]
5. Fernberg P, Odenbro Å, Bellocco R et al. Tobacco use, body mass index and the risk of malignant lymphomas – A nationwide cohort study in Sweden. Int J Cancer 2006: 118: 2298–2302.[CrossRef][ISI][Medline]
6. Morton LM, Zheng T, Holford TR et al. Alcohol consumption and risk of non-Hodgkin lymphoma: a pooled analysis. Lancet Oncol 2005; 6: 469–476.[CrossRef][ISI][Medline]
7. Freedman DS, Tolbert PE, Coates R et al. Relation of cigarette smoking to non-Hodgkin's lymphoma among middle-aged men. Am J Epidemiol 1998; 148: 833–841.
8. Cartwright RA, McKinney PA, O'Brien C et al. Non-Hodgkin's lymphoma: case control epidemiological study in Yorkshire. Leuk Res 1988; 12: 81–88.[CrossRef][ISI][Medline]
9. De Stefani E, Fierro L, Barrios E, Ronco A. Tobacco, alcohol, diet and risk of non-Hodgkin's lymphoma: a case-control study in Uruguay. Leuk Res 1998; 22: 445–452.[CrossRef][ISI][Medline]
10. Franceschi S, Serraino D, Carbone A et al. Dietary factors and non-Hodgkin's lymphoma: a case-control study in the Northeastern part of Italy. Nutr Cancer 1989; 12: 333–341.[ISI][Medline]
11. Tavani A, Pregnolato A, Negri E et al. Diet and risk of lymphoid neoplasms and soft tissue sarcomas. Nutr Cancer 1997; 27: 256–260.[ISI][Medline]
12. Stagnaro E, Ramazzotti V, Crosignani P et al. Smoking and hematolymphopoietic malignancies. Cancer Causes Control 2001; 12: 325–334.[CrossRef][ISI][Medline]
13. Seniori Costantini A, Miligi L, Kriebel D et al. A multicenter case-control study in Italy on hematolymphopoietic neoplasms and occupation. Epidemiology 2001; 12: 78–87.[CrossRef][ISI][Medline]
14. The Non-Hodgkin's Lymphoma Pathologic Classification Project. National Cancer Institute sponsored study of classifications of non-Hodgkin's lymphomas: summary and description of a Working Formulation for clinical usage. Cancer 1982; 49: 2112–2135.[CrossRef][ISI][Medline]
15. Collett D. Modelling survival data in medical research. U.S.A.: Chapman & Hall/CRC, 2003.
16. Sant M, Aareleid T, Berrino, F et al. EUROCARE-3: survival of cancer patients diagnosed 1990–94–Results and commentary. Ann Oncol 2003; 14: 61–118.[CrossRef]
17. The Non-Hodgkin's Lymphoma Classification Project. A clinical evaluation of the International Lymphoma Study Group Classification of non-Hodgkin's Lymphoma. Blood 1997; 89: 3909–3918.
18. Vineis P, Alavanja M, Buffler P et al. Tobacco and cancer: recent epidemiological evidence. J Natl Cancer Inst 2004; 96: 99–106.
19. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Tobacco smoke and involuntary smoking. IARC Monogr Eval Carcinog Risks Hum 2004; 83: 1–1438.[Medline]
20. Bell DA, Liu Y, Cortopassi GA. Occurrence of bcl-2 oncogene translocation with increases frequency in the peripheral blood of heavy smokers. J Natl Cancer Inst 1995; 87: 223–224.
21. Berglund M, Thunberg U, Amini RM et al. Evaluation of immunophenoptype in diffuse large B-cell lymphoma and its impact on prognosis. Mod Pathol 2005; 18: 1113–1120.[CrossRef][ISI][Medline]
22. Nieder C, Petersen S, Pertersen C, Thames HD. The challenge of p53 as prognostic and predictive factor in non-Hodgkin's lymphoma. Am J Hematol 2001; 80: 2–8.[CrossRef]
23. Sopori ML, Kozak W. Immunomodulatory effect of cigarette smoke. J Neuroimmunol 1998; 83: 148–156.[CrossRef][ISI][Medline]
24. Kalra R, Singh SP, Savage SM. Effects of cigarette smoke on immune response: chronic exposure to cigarette smoke impairs antigen-mediated signaling in T Cells and depletes IP3-sensitive Ca2+ stores. J Pharmacol Exp Ther 2000; 293: 166–171.
25. Andersen P, Pedersen OF, Bach B, Bonde GJ. Serum antibodies and immunoglobulins in smokers and nonsmokers. J Clin Exp Immunol 1982; 47: 467–73.[ISI][Medline]
26. Schwartz J, Weiss ST. Host and environmental factors influencing the peripheral blood leukocyte count. Am J Epidemiol 1991; 134: 1402–1409.
27. Brooks PJ, Theruvathu JA. DNA adducts from acetaldehide: implications for alcohol-related carcinogenensis. Alcohol 2005; 35: 187–193.[CrossRef][ISI][Medline]
28. Diaz L, Montero A, Gonzalez-Gross M. Influence of alcohol consumption on immunological status: a review. Eur J Clin Nutr 2002; 56: S50–S53.
29. Patrick DL, Cheadle A, Thompson DC et al. The validity of self-reported smoking: a review and meta-analysis. Am J Public Health 1994; 84: 1086–1093.
30. Clarke CA, Glaser SL, Dorfman RF et al. Expert review of non-Hodgkin's lymphomas in a population-based cancer registry: reliability of diagnosis and subtype classifications. Cancer Epidemiol Biomarkers Prev 2004; 13: 138–143.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||