Critique 123: Alcohol consumption and lymphoid and myeloid neoplasms — 5 September 2013
Heinen MM, Verhage BAJ, Schouten LJ, Goldbohm RA, Schouten HC, van den Brandt PA. Alcohol consumption and risk of lymphoid and myeloid neoplasms: Results of the Netherlands cohort study. Int J Cancer 2013l133:1701–1713.
Results from epidemiological studies suggest that alcohol drinkers have a decreased risk of lymphoid neoplasms, whereas results for myeloid neoplasms are inconsistent. However, most of these studies have used retrospective data. We examined prospectively whether alcohol consumption decreases the risk of both lymphoid and myeloid neoplasms, including most common subtypes. Moreover, we investigated whether this decreased risk is due to ethanol or other contents of specific alcoholic beverages (i.e., beer, wine and liquor).
The Netherlands cohort study consisted of 120,852 individuals who completed a baseline questionnaire in 1986. After 17.3 years of follow-up, 1,375 cases of lymphoid and 245 cases of myeloid neoplasms with complete exposure information were available for analysis. Compared with abstinence, we observed for plasma cell neoplasms hazard rate ratios (HR) of 1.66 (95% confidence interval (CI), 1.21–2.29), 1.63 (95% CI, 1.17–2.27), 1.11 (95% CI, 0.75–1.64) and 0.85 (95% CI, 0.51–1.42) with daily ethanol consumption of 0.1–<5, 5–<15, 15–<30 and ≥30 g, respectively. A similar pattern was observed for chronic lymphocytic leukemia/small lymphocytic lymphoma. No associations were observed for other subtypes and for myeloid neoplasms. When results were analyzed by beverage type, no clear associations were observed.
In conclusion, our study did not show an inverse association between alcohol consumption and lymphoid neoplasms. Also, no inverse association was observed with myeloid neoplasms. If any association between alcohol consumption and lymphoid neoplasms exists, our study suggests an increased risk rather than a decreased risk.
Background: Many prospective studies have shown that moderate drinkers are at lower risk of certain lymphoid cancers. For example, the Million Women’s Study in the UK found that alcohol consumption showed a significant inverse association with the occurrence of non-Hodgkin lymphoma (Allen NE, Beral V, Casabonne D, Kan SW, Reeves GK, Brown A, Green J, for the Million Women Study Collaborators. Moderate alcohol intake and cancer incidence in women. J Natl Cancer Inst 2009;101:296-305). In a more recent publication from this study, investigators reported on 9,162 incident cases of haematological malignancy, including 7,047 lymphoid and 2,072 myeloid cancers. They conclude: “Among predominantly moderate alcohol drinkers, higher intake was associated with lower risk of lymphoid malignancies,” but they did not find a protective effect on the risk of myeloid tumors such as acute myeloid leukemia (Kroll ME, Murphy F, Pirie K, Reeves GK, Green J, Beral V, for the Million Women Study Collaborators. Alcohol drinking, tobacco smoking and subtypes of haematological malignancy in the UK Million Women Study. British Journal of Cancer 2012;107:879–887).
Overall, analyses from the Million Women Study, with much larger numbers of cases than the present analysis, indicate that none of the types of hematological cancer showed an increase in risk with alcohol consumption. Instead, in comparison with non-drinkers, the risk of lymphoid tumors was reduced among consumers of up to 3 drinks/week (the referent group), and reduced further among those consuming larger amounts of alcohol. The risk of myeloid tumors was not affected by alcohol intake. (A critique of the paper by Kroll et al is available as review # 095 on the Forum web-site, www.bu.edu/alcohol-forum).
The authors of the present study list the reference to, but do not discuss, a paper by Gapstur et al from 2012 that had almost 2,000 cases of Non-Hodgkin Lymphoma (Gapstur SM, Diver WR, McCullough ML, Teras LR, Thun MJ, Patel AV. Alcohol Intake and the Incidence of Non-Hodgkin Lymphoid Neoplasms in the Cancer Prevention Study II Nutrition Cohort. Am J Epidemiol 2012;176:60-69). That paper concluded: “In summary, findings from the prospective study presented here support the hypothesis that alcohol intake might be associated with a reduced risk of NHL. Furthermore, they indicate no statistically significant association with former intake and no difference in associations between men and women, between ever and never smokers, among NHL subtypes, or by beverage type.”
Specific comments on the present study: The present study apparently gives results that are different from many previous epidemiologic studies, in that he reports no decrease in lymphoid malignancies with moderate drinking. There are a number of strengths to the study: a very large group of subjects with more than 17 years of follow up and essentially full ascertainment of cases of cancer. There were 1,375 cases of lymphoid and 245 cases of myeloid neoplasms occurring during follow up.
However, there are some questionable aspects of the study and the authors’ interpretation of the results. The assessment of alcohol consumption was based only on a baseline questionnaire , but with no data available during the long follow-up period. The subjects were 55-69 years old at baseline, but the authors claim that by having a question at baseline about alcohol intake 5 years previously, they were able to identify lifetime abstainers (which is doubtful).
Questions about the analyses: Forum members had a number of questions about the analyses in this paper. The investigators used a sub-cohort of “control subjects” (5,000 subjects chosen from their large number of total subjects). These control subjects had some very large differences from the subjects developing cancer, including having a much lower occurrence of a positive family history of hematologic malignancies. However, the authors only present data adjusted by age and sex, stating that “none of the potential confounding variables were associated with the neoplasms studied and with ethanol intake and changed the risk estimate by at least 10%,” so they ignored them. They go on to comment in the text: “Inclusion of confounding variables in the Cox regression models yielded very similar results (data not shown).”
Another disturbing finding is that in most of the statistically significant “positive” associations they report, there was no evidence of an increasing effect with increasing amounts of alcohol (dose-response relation). In fact, there were a number of inverse associations between alcohol intake and cancer for some types of cancer. For example, in one important set of analyses, the authors state “When the abstainers were excluded from the analyses and moderate alcohol consumers (0.1-<5 g/day) were used as the reference group, the inverse trends for the lymphoid neoplasms overall and the subtype PCN became even more significant (p for trend = 0.004 and 0.001, respectively).” Despite these inconsistencies, they conclude that their study suggests an increased risk rather than a decreased risk of these cancers to be associated with alcohol consumption.
Reviewer Skovenborg stated: “The results of this large, prospective cohort study are inconclusive and the inverse dose-response curve is indeed a very strange finding for a proposed risk factor. Moderate consumption is defined as drinking >10 – <30 g alcohol per day, and the increased risks of hematologic neoplasms were found in the low drinking categories: 0.1 – <5, and 5 – <15 g/day. Even so, the authors state that a statistically significant increased risks of CLL/SLL and PCN was found for moderate alcohol consumption.”
Forum reviewer Zhang was also puzzled by some of the results of this study. “Overall, statistical methods used to analyze data seem to be appropriate. However, to keep ‘consistency’ they tested linear trend for all dose response relationships, even though their own findings indicated such linear trend may not be present. For example there is no linear relationship between alcohol consumption and risk of PCN, but p for trend was =0.03. I am wondering how the authors could obtain such result (hazard ratios for each increased alcohol consumption groups were 1.00, 1.66, 1.63, 1.11, 0.85, p=0.03). According to the spline curve, the risk of cancer increases as alcohol consumption increases up to 6 g/d, and then the risk decreases. In fact, those who consumed more than 20g/d had lower risk than those abstainers. These findings are difficult to interpret.”
Forum member Waterhouse also had concerns about this study. “The authors essentially demonstrate an inverse dose response curve, which is very difficult to understand. In addition, their result showing that the very large population of the lightest drinkers (from 0.1 to no more than 5 g/day) had a very high RR of 1.66 for plasma cell neoplasms, the highest in the study. Such a result is confusing, at best. (Perhaps the light drinkers included a lot of weekly binge drinkers and that affected the result?) They seem to provide little explanation for these confusing results.”
Given the rather consistent findings in other very large studies of hematological malignancies that lymphoid types may be reduced by alcohol, and the inconsistencies in the results of the present analyses, it is difficult to conclude that the present paper should change our current interpretation of the association between alcohol and these types of cancer. Forum members look forward to reports from other prospective studies on this topic.
Many prospective studies have shown that moderate drinkers are at lower risk of certain types of lymphoid cancer. For example, a report in 2009 from the Million Women’s Study in the UK found that alcohol consumption showed a significant inverse association with the occurrence of non-Hodgkin lymphoma (NHL). Further, a 2012 paper from that study based on 9,162 incident cases of haematological malignancy, including 7,047 lymphoid and 2,072 myeloid cancers, concluded: “Among predominantly moderate alcohol drinkers, higher intake was associated with lower risk of lymphoid malignancies.” Those investigators did not find a significant effect of alcohol on the risk of myeloid tumors, such as acute myeloid leukemia. Another 2012 paper based on almost 2,000 cases of Non-Hodgkin Lymphoma concluded: “In summary, findings from the prospective study presented here support the hypothesis that alcohol intake might be associated with a reduced risk of NHL.”
The authors of the present paper from the Netherlands, based on 17.3 years of follow up with 1,375 cases of lymphoid and 245 cases of myeloid neoplasms, did not find a statistically significant reduction in the risk of lymphoid cancers, and the authors suggest: “If any association between alcohol consumption and lymphoid neoplasms exists, our study suggests an increased risk rather than a decreased risk.”
While reasons for this difference in results of this study when compared with other recent studies may be used to better understand the association between alcohol and such cancers, Forum reviewers were concerned about a number of aspects of this study. For example, the authors had data on alcohol consumption only at baseline for this very long (17+ year) follow up, so changes (either increases or decreases in intake) were not known. Further, the key relations reported were not adjusted for a large number of potentially confounding variables, such as a positive family history of hematological cancer. (The authors state that they carried out such multivariable analyses but do not present the data.) Of even more concern was the apparent inverse effect of alcohol on cancer risk for a number of types of tumors when dose-response relations were shown.
Given the rather consistent findings in other very large studies of hematological malignancies that the risk of many types of lymphoid cancer may be reduced by alcohol, and inconsistencies in the results of the present analyses, it is difficult to conclude that the present paper should change our current interpretation of the association between alcohol and these types of cancer.
Forum members look forward to reports from other prospective studies on this topic. For the present, however, the overwhelming scientific evidence suggests that moderate alcohol consumption is associated with a decrease in the risk of many types of lymphoid malignancies, but has little effect on myeloid cancers.
* * *
Comments on this critique were provided by the following members of the International Scientific Forum on Alcohol Research:
Harvey Finkel, MD, Hematology/Oncology,
Creina Stockley, PhD, MBA, Clinical Pharmacology, Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia
Yuqing Zhang, MD, DSc, Epidemiology, Boston University School of Medicine, Boston, MA, USA
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA