Dietary One-Carbon Nutrient Intake and Risk of Lymphoid and Myeloid Neoplasms: Results of the Netherlands Cohort Study Free

Hematologic malignancies are a heterogeneous group of neoplasms that originate from lymphoid and myeloid cells. They account for 7.4% of cancers in males and 6.4% in females worldwide (1). So far, little is known about the causes and only few factors have been linked to lymphoid neoplasms, including age, gender, primary/inherited immune deficiencies, a number of viruses including the human immunodeficiency virus, and several autoimmune diseases (2). Risk factors for myeloid neoplasms include age, gender, genetic abnormalities, family history, and exposures to radiation and benzene (3). Stratification for subtypes of lymphoid and myeloid neoplasms in previous studies suggested etiologic heterogeneity (4, 5) and, therefore, further research is warranted in which these subtypes with sufficient numbers can be examined.

Previous epidemiologic research suggests a protective role of dietary one-carbon nutrients in colorectal carcinogenesis (6, 7) and perhaps other cancers, such as breast, ovarian, and gastric (7–10). Furthermore, research has suggested that alcohol influences cancer risk by antagonizing the one-carbon metabolism (11). However, research also shows that folate may play a dual role in cancer development: supplementation may provide protection early in carcinogenesis in individuals with a low folate status, yet supplementation late in carcinogenesis and potentially at very high intakes may promote carcinogenesis (12–14). Dietary folate consists of monoglutamate and polyglutamate folate species. Many studies have determined that the relative bioavailability of polyglutamates versus corresponding monoglutamates ranges from 50% to 100% (15), therefore, beneficial effects of folate intake might be stronger for folate monoglutamates than for folate polyglutamates.

So far, data on the association between one-carbon nutrients and the risk of lymphoid neoplasms are limited to two cohort (16, 17) and four case–control studies (18–21). One cohort study among U.S. women observed no association between folate and non-Hodgkin lymphoma (NHL) risk (17). A cohort study among male smokers observed a borderline significantly decreased risk between vitamin B12 and NHL risk, but no associations with other dietary one-carbon nutrients and no associations among lymphoma subtypes and multiple myeloma (16). As for the case–control studies, significantly decreased risks were observed for folate intake with NHL overall (18, 19), with the subtype diffuse large B-cell lymphoma (DLBCL; refs. 19, 20) and with marginal zone lymphoma (19). Decreased risks were also observed for higher vitamin B6 intake with NHL overall (20), with DLBCL (20) and with marginal zone lymphoma (19). For methionine intake, decreased risks were observed with NHL overall (19, 20), with DLBCL (20) and with follicular lymphoma (19, 20). One hospital-based case–control study did only observe associations between one-carbon nutrients and NHL risk when the analyses were stratified by current drinking status (21). Significant decreased risks were observed for folate intake among the abstainers and former drinkers, whereas no associations were observed among current drinkers (21). Another case–control study did, however, not observe any modifying effects of alcohol on NHL risk (20). As far as the researchers know, no epidemiologic studies have been conducted on the relation between one-carbon nutrients and risk of myeloid neoplasms.

We investigated the relations between dietary one-carbon nutrients [i.e., total folate, poly- and monoglutamates, riboflavin (vitamin B2), vitamin B6, and methionine] and risk of lymphoid and myeloid neoplasms, including the most common subtypes, in a large prospective cohort study of men and women in the Netherlands. Furthermore, we examined whether the effect of dietary folate was modified by alcohol intake.

The study design of the Netherlands Cohort Study (NLCS) has been reported in detail elsewhere (22). Briefly, the NLCS was initiated in September 1986 and included 58,279 men and 62,573 women ages 55 to 69 years at the beginning of the study, originating from 204 municipalities with computerized population registries. A self-administered questionnaire on daily dietary habits, lifestyle factors, and other potential risk factors for cancer was completed at baseline. For reasons of efficiency in questionnaire processing and follow-up, the case-cohort approach (23) was used. Incident cases were derived from the entire cohort, whereas the person-years at risk were estimated from a random sample of 5,000 subjects. This subcohort was chosen immediately after baseline and followed up for vital status information. The entire cohort is being monitored for cancer occurrence by annual record linkage to the Netherlands Cancer Registry and the Netherlands Pathology Registry (24, 25). A total of 17.3 years of follow-up (baseline to December 2003; mean follow-up = 13.7 years) were used for the current analysis. Only one subcohort member was lost to follow-up. The completeness of cancer follow-up was estimated to be >96% (26). All prevalent cancer cases at baseline other than skin cancer were excluded (n = 226), leaving 4,774 subcohort members (Fig. 1).

Cases were defined as participants with an incident, histologically verified diagnosis of lymphoid neoplasms (n = 1,280) or myeloid neoplasms (n = 222). Histology is coded by the Netherlands Cancer Registry using the International Classification of Diseases for Oncology (27). Using the histology codes provided by the cancer registries, we subdivided the lymphoid neoplasms into categories based on the hierarchical groupings of the International Lymphoma Epidemiology Consortium (InterLymph) nested classification (Table 1; ref. 28). This classification is based on the World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissues (29) and the International Classification of Diseases for Oncology, Third Edition (27). About the myeloid neoplasms, we grouped these malignancies into categories defined by the WHO classification (Table 1; ref. 29). For cases that could not be assigned to a specific category (e.g., the “not otherwise specified” categories), the summary of the pathology report (received from the Netherlands Pathology Registry PALGA; ref. 24) was inspected and, if possible, an appropriate category was assigned. This was the case for 66.4% (n = 75) for NHL, B cell, NOS; 55.6% (n = 10) for NHL, T cell, NOS; 81.1% (n = 73) for NHL, unknown lineage; 77.8% (n = 35) for lymphoid neoplasms, NOS; and 100% (n = 2) for leukemia, NOS.

Subjects with incomplete or inconsistent dietary data were excluded from analysis. Details are given elsewhere (30). Figure 1 shows the selection and exclusion steps that resulted in the final numbers of cases and subcohort members that were available for analysis. The NLCS has been approved by the Institutional Review Boards of the TNO Nutrition and Food Research Institute (Zeist, the Netherlands) and Maastricht University (Maastricht, the Netherlands).

The dietary section of the questionnaire was a 150-item semiquantitative food-frequency questionnaire (FFQ), which concentrated on the habitual consumption during the year preceding the start of the study. Questionnaire data were key-entered and processed for all incident cases in the cohort and subcohort members in a standardized manner blinded with respect to case/subcohort status. This was done to minimize observer bias in coding and interpretation of the data. Folate intake was calculated using data from a validated liquid chromatography trienzyme method used to analyze the 125 most important Dutch food items contributing to folate intake (31). Daily mean intakes of all other relevant nutrients were calculated by summing the multiplied frequencies and portion sizes of all food items with their tabulated nutrient contents from the Dutch Food Composition Table (32). Data on dietary supplement use were also obtained via the FFQ. However, the use of B-vitamin and multivitamin supplements was low (7.3% and 4.6% in subcohort members, respectively). Moreover, due to legislative restrictions in the Netherlands, it was not allowed until 1994 to use folic acid in vitamin supplements (33). Therefore, folic acid and vitamin B supplement use most likely plays a very minor role in our study population, and supplement use was not further accounted for in the analyses. The FFQ has been validated and tested for reproducibility (30, 34). Pearson correlation coefficients for nutrient intake evaluated by a 9-day diet record and the questionnaire ranged from 0.6 to 0.8 for most nutrients.

Cox proportional hazards models were used to estimate incidence HRs and corresponding 95% confidence intervals (CI). The total person-years at risk estimated from the subcohort were used in the analyses (35). SEs were estimated using a robust covariance matrix estimator to account for increased variance due to sampling from the cohort (36). The proportional hazards assumption was tested using the scaled Schoenfeld residuals (37).

All analyses were conducted for both sexes combined and separately for men and women. Furthermore, interactions on a multiplicative scale between sex and the exposure variables were tested for lymphoid and myeloid neoplasms. Because of low case numbers (n < 100), we did not conduct stratified analyses on sex for the subtypes follicular lymphoma and lymphoplasmacytic lymphoma/Waldenström (LL/WM). HRs were estimated for quintiles of intake (with the lowest quintile of intake regarded as the reference group) based on the gender-specific distribution in the subcohort.

Variables examined as potential confounders included body mass index, height, energy intake, intake of vegetables, fruit, red meat, dairy, alcohol, fat, and protein, level of education, nonoccupational physical activity, smoking, family history of hematologic malignancies, history of chronic bowel irritation, and rheumatoid arthritis. These potential confounding variables were regarded as confounders if they (i) were associated with lymphoid and myeloid neoplasms and with the intake of the exposure variables and (ii) changed the age- and sex-adjusted risk estimates by at least 10% (using a backwards stepwise procedure). This resulted in a multivariable-adjusted model including age at baseline (years), gender, cigarette smoking (current smoking: yes/no; number of cigarettes smoked per day; number of years of smoking), height (cm), family history of hematologic malignancies (yes/no), level of education (primary school or lower vocational school/intermediate vocational school or high school/higher vocational school or college), and intake of energy (kcal/day) and alcohol (abstainer, 0.1–4, 5–14, 15–29, and ≥30 g/day). For the analyses of monoglutamates, polyglutamates were included simultaneously and vice versa. For each analysis, trends were evaluated with the Wald test by assigning participants the median value for each level of the categorical exposure variable among the subcohort members, and this variable was entered as a continuous term in the Cox regression model. To evaluate whether early symptoms of disease before diagnosis could have influenced the results, early cases (diagnosed within 2 years after baseline) were excluded in additional analyses. To investigate whether alcohol modified the effect of folate on the risk of lymphoid and myeloid neoplasms, multiplicative interaction terms were used in the regression models and stratum-specific HR estimates were examined. Folate intake was analyzed in tertiles, and three categories of alcohol consumption (abstainer, 0.1-<30; ≥30 g/day) were used for this analysis. All analyses were performed using the STATA statistical software package (Intercooled STATA, version 10; Stata Corporation). All P values were based on two-sided tests and considered statistically significant if <0.05.

In the present study, the analysis was performed for lymphoid (n = 1,280) and myeloid neoplasms (n = 222) overall and for specific subtypes with a sufficient number of incident cases (n ≥ 70, as used in a previous analysis within the NLCS; ref. 38; Table 1). The subtypes included were chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL; n = 213), LL/WM (n = 89), follicular lymphoma (n = 96), DLBCL (n = 248), and plasma cell neoplasms (PCN; n = 343) for lymphoid neoplasms and acute myelogenous leukemia (AML; n = 157) for myeloid neoplasms.

In Tables 2 and 3, baseline characteristics are presented for cases (overall and by subtype) and subcohort members. Most characteristics did not differ noticeably between cases and subcohort members, but there were more men among cases than among subcohort members. In addition, there were more subjects with a family history of hematologic malignancies among cases than among subcohort members, especially among follicular lymphoma and AML cases (2.7% among subcohort members, 7.3% among follicular lymphoma, and 6.4% among AML cases).

Table 4 shows the multivariable-adjusted HRs and 95% CIs for the association between one-carbon nutrients and lymphoid and myeloid neoplasms, including subtypes. Intakes of folate, methionine, and riboflavin were not associated with lymphoid and myeloid neoplasms. For vitamin B6, a statistically significantly increased myeloid neoplasms risk was observed (HR = 1.87; 95% CI, 1.08–3.25 for highest vs. lowest quintile of intake, P_trend = 0.02; Table 4). Analysis by lymphoid and myeloid subtypes indicated significantly increased PCN risks for folate (HR = 1.73; 95% CI, 1.14–2.61 for highest vs. lowest quintile of intake) and vitamin B6 intake (HR = 1.66; 95% CI, 1.03–2.67) and a significantly lower DLBCL risk for methionine intake (HR = 0.56; 95% CI, 0.33–0.97). No clear associations were observed for other subtypes.

When examining the associations between the risk for lymphoid and myeloid neoplasms and intake of mono- and polyglutamates, no clear associations were observed for lymphoid and myeloid neoplasms overall and for most of the subtypes (Supplementary Table S1). For LL/WM, decreased risks were observed for intake of monoglutamates. However, only the second and third quintile compared with the first quintile reached statistical significance.

As shown in Table 5, no evidence for an interaction between alcohol consumption and folate intake was observed (P_interaction>0.05). Although, for individuals who used ≥30 g of alcohol daily and were in the second tertile of folate intake, we observed a decreased association for lymphoid neoplasms (HR = 0.60; 95% CI, 0.36–0.98; Table 4). For all other categories, no associations were observed for both lymphoid neoplasms and myeloid neoplasms.

When we stratified the analysis for lymphoid and myeloid neoplasms risk by sex, results on methionine, riboflavin, vitamin B6, mono-, and polyglutamates were similar for men and women and the tests for interactions between these nutrients and sex were not significant (P > 0.05; data not shown). For folate intake, we observed statistically significantly increased lymphoid neoplasms risks among women (HR = 1.38, 95% CI, 1.00–1.92; HR = 1.45, 95% CI, 1.05–2.00; HR = 1.14, 95% CI, 0.81–1.62; and HR = 1.45, 95% CI, 1.01–2.07 for quintiles 2–5 vs. the first quintile, respectively). Only for the lymphoma subtype PCN, similar increased risks were observed (data not shown). No associations with folate intake were observed among men. After excluding the first 2 years of follow-up (85 lymphoid neoplasm cases and 15 myeloid neoplasm cases), some of the statistically significantly increased and decreased risks became nonsignificant. This included the decreased DLBCL risk for methionine (HR = 0.59; 95% CI, 0.34–1.02 for highest vs. lowest quintile) and the increased PCN risk for vitamin B6 (HR = 1.58; 95% CI, 0.97–2.59). To investigate this into more detail, a test for interaction between the exposure of interest and time was calculated and the remaining follow-up period was stratified into three periods (≥2–<7, ≥7–<12, and ≥12 years). The interactions with time were, however, not significant and no significant results were observed for each of the periods (results not shown).

Overall, most one-carbon nutrients were not associated with lymphoid or myeloid neoplasms in the present prospective study. For vitamin B6, however, a statistically significantly increased myeloid neoplasms risk was observed (highest vs. lowest quintile of intake: HR = 1.87). Furthermore, analysis by subtype showed significantly increased PCN risks for folate (HR = 1.73) and vitamin B6 intake (HR = 1.66) and a decreased DLBCL risk for methionine intake (HR = 0.56). However, some of these risk estimates became statistically nonsignificant after excluding the first 2 years of follow-up.

The few epidemiologic studies investigating the relation between folate, methionine, riboflavin, or vitamin B6 and risk of lymphoid neoplasms have shown inconsistent results so far. A cohort study among male smokers observed no associations of folate, methionine, riboflavin, and vitamin B6 with NHL risk overall, with subtypes, and with multiple myeloma (16). Two case–control studies, on the other hand, observed that higher intakes of folate (19, 20), vitamin B6 (20), and methionine (20) were inversely associated with DLBCL risk. Decreased risks were also observed for follicular lymphoma and methionine intake (19, 20). Only one population-based case–control study among Connecticut women observed a statistically significant increased risk of DLBCL with higher riboflavin intake (OR = 1.74; 95% CI, 1.02–2.96, highest quartile vs. lowest quartile; ref. 19).

About the modifying effects of alcohol on folate intake, an Italian hospital-based case–control study observed significant decreased risks for NHL and folate intake among the subgroup of abstainers and former drinkers, whereas no associations were observed among current drinkers (21). A case–control study among U.S. adults did not observe any interaction between alcohol and folate on NHL risk (20), which is in agreement with our data. The relation between alcohol and risk of lymphoid and myeloid neoplasms has already been investigated within the NLCS, in which no clear relations were observed (39).

Some evidences on the effects of one-carbon nutrients on carcinogenesis suggest a protective effect of these nutrients (7, 40). Deficiencies in folate and methionine can lead to an impaired immune response (41), which is a risk factor for lymphoid neoplasm. In addition, deficiencies in folate and other nutrients contributing to the one-carbon metabolism can impair DNA methylation (40). For example, hypermethylation of suppressor gene promoters has been associated with transcriptional silencing, which may be involved in promoting hematopoietic malignancies (42). This suggest that higher dietary intake of these nutrients could confer a lower risk of lymphoid neoplasm. However, the few statistically significant associations observed in the current study were positive rather than inverse. Our findings are supported by previous research showing that high intakes of folate are associated with an increased cancer risk (43, 44). Folate supplementation may promote the progression of neoplastic lesions as folate plays an important role in DNA synthesis and replication, especially in rapidly dividing cells such as neoplastic cells. Interruption of folate metabolism causes ineffective DNA synthesis, resulting in inhibition of tumor growth (12, 13). This serves as the basis for treatment of lymphoid malignancies with antifolate chemotherapeutic agents, drugs that interrupt normal cellular metabolism and slow growth or induce cell death (45).

Changes in our understanding of hematologic malignancies have resulted in the evolution of several classification schemes over the last several decades. In 2007, the Pathology Working Group of the InterLymph has developed a hierarchical scheme for the classification of lymphoid neoplasms and, in addition, provided a translation from previous classifications (28). This scheme will facilitate subtype-specific analyses to the most detailed extent possible, which is critical for comparing subtypes-specific data among epidemiologic studies. In the current study, this hierarchical scheme was used. The use of different classifications in previous studies makes meaningful comparison of results between epidemiologic studies difficult and might explain the observed differences between studies.

Because of the large number of comparisons that were made in the present study, some of the observed statistical significant associations would be expected to arise due to chance alone. Furthermore, excluding the first 2 years of follow-up slightly attenuated some of our findings. Therefore, our results must be regarded with some caution and need replication. Some of the heterogeneity among previous studies may partly be due to the fact that case–control studies are notoriously prone to recall and selection bias, particularly when the exposure under study is diet (46). Also, reasons for nonparticipation may have differed between cases and controls and self-selection of systematically different controls may have taken place, which could have led to spurious findings. Moreover, results from two large population-based case–control studies that investigated gene-nutrient interactions among determinants of one-carbon metabolism on the risk of NHL, suggested that the NHL risk associated with one-carbon metabolizing pathway genes seems to be contingent upon the dietary intakes of nutrients related to one-carbon metabolism (47, 48). These intriguing new results require replication in future epidemiologic studies.

This was the first study to look at the relation of mono- and polyglutamates with lymphoid and myeloid neoplasms. No difference in risk was observed, although one expects to see a difference with the intake of mono- and polyglutamates, because the relative bioavailability of polyglutamates versus their corresponding monoglutamates might be as low as 50% (15). However, more recent studies have shown that there was no evidence that foods that have a greater percentage of polyglutamates were less bioavailable than those with less (49). For example, yeast that contains 100% polyglutamates had a higher bioavailability than spinach (with 50% polyglutamates), showing relative bioavailability of 31% to 44% for spinach and 45% to 62% for yeast compared with folic acid (100%). So, it remains to be seen whether there is a difference to be expected in risks between mono- and polyglutamates. In the current study, data were also available for several folate vitamers. However, these are a mix of mono- and polyglutamates and therefore not presented. Nevertheless, when the associations between these separate folate vitamers and risk of lymphoid and myeloid neoplasms and their subtypes were examined, no clear associations or patterns were observed (data not shown).

One of the limitations of this study is the use of a single measure of dietary intake in the current study that may not have been representative of the dietary habits of the study participants over the course of follow-up. The FFQ was tested for reproducibility by Goldbohm and colleagues (34) who concluded that the single measurement of intake of diet in the NLCS can characterize dietary habits for a period of at least 5 years. Moreover, subjects included in the current study ages 55 to 69 years at baseline and these elderly people tend to have more stable dietary habits compared with younger individuals (22). Another limitation is that the number of cases in some of the subgroups was low, leading to limited power to explore the relation with intakes of one-carbon nutrients in some subgroups and even precluded analyses for some other subgroups. Although, the current study has the largest case series available to date that investigated these relations.

Strengths of this study include the prospective design, which avoids the possibility of biased recall. Also, differential follow-up is unlikely to have made material contribution to our findings, as completeness of follow-up was high (26). Other strengths include detailed information on potential confounders and the use of a compiled folate database, with food folate values, which was established using a validated trienzyme high-performance liquid chromatography method (31).

Overall, we observed a few statistically significant positive associations and a decreased DLBCL risk for methionine intake only. However, some of these observed statistical significant associations would be expected to arise due to chance alone and some of the risk estimates became nonsignificant when excluding early cases. Therefore, we conclude that there is no association between one-carbon nutrient intake and risk of lymphoid and myeloid neoplasms in the current study. Furthermore, there was no evidence for an interaction between alcohol and folate on the risk of these neoplasms.

No potential conflicts of interest were disclosed.

Conception and design: P.A. van den Brandt, L.J. Schouten, R.A. Goldbohm

Development of methodology: P.A. van den Brandt, R.A. Goldbohm

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.A. van den Brandt, L.J. Schouten, R.A. Goldbohm

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.M. Heinen, P.A. van den Brandt, H.C. Schouten, B.A.J. Verhage

Writing, review, and/or revision of the manuscript: M.M. Heinen, P.A. van den Brandt, L.J. Schouten, R.A. Goldbohm, H.C. Schouten, B.A.J. Verhage

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.M. Heinen, B.A.J. Verhage

Study supervision: P.A. van den Brandt

The authors thank the participants of this study, the Netherlands Cancer Registry, the Netherlands nationwide registry of pathology (PALGA), Drs. A. Volovics and A. Kester for statistical advice, S. van de Crommert, H. Brants, J. Nelissen, C. de Zwart, M. Moll, and A. Pisters for assistance, and H. van Montfort, T. van Moergastel, L. van den Bosch, E. Dutman, R. Schmeitz, J. Berben, and R. Meijer for programming assistance.

This work was financially supported by the European Foundation for Alcohol Research (former European Research Advisory Board; ERAB).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.