Kang JH, Loomis SJ, Wiggs JL, Willett WC, Pasquale LR. A Prospective Study of Folate, Vitamin B6, and Vitamin B12 Intake in Relation to Exfoliation Glaucoma or Suspected Exfoliation Glaucoma. JAMA Ophthalmol. 2014;132(5):549-559. doi:10.1001/jamaophthalmol.2014.100
Effective strategies for primary prevention are lacking for exfoliation glaucoma (EG), which is the most common type of secondary glaucoma.
To examine the association between B vitamin intake and EG or suspected EG (EG/SEG) risk.
Design, Setting, and Participants
National prospective cohort study using more than 20 years of follow-up data from the Nurses’ Health Study (all female registered nurses) and the Health Professionals Follow-up Study (all male health professionals) from June 1, 1980, to May 31, 2010 (Nurses’ Health Study) and January 1, 1986, to December 31, 2010 (Health Professionals Follow-up Study). We included a subset of 78 980 Nurses’ Health Study women and 41 221 Health Professionals Follow-up Study men who were 40 years or older, free of glaucoma, had completed diet questionnaires, and reported eye examinations (follow-up rate, >85%).
Cumulatively updated intake of B vitamins (folate, vitamin B6, and vitamin B12) as ascertained by repeated administration of validated questionnaires.
Main Outcomes and Measures
Incident cases of EG/SEG, totaling 399 (329 women and 70 men), were first identified with the questionnaires and were subsequently confirmed with medical records. Multivariable relative risks for EG/SEG were calculated in each cohort and then pooled with meta-analysis.
Vitamin B6 and vitamin B12 intake was not associated with EG/SEG risk in pooled analyses (P = .52 and P = .99 for linear trend, respectively). However, a suggestive trend of a reduced risk was observed with higher intake of folate: compared with the lowest quintile of cumulatively averaged updated total folate intake, the multivariable relative risk for EG/SEG for the highest quintile (≥654 μg/d) was 0.75 (95% CI, 0.54-1.04; P = .02 for linear trend). These results were not materially altered after adjustment for vitamin B6 and vitamin B12 intake. An association was observed for supplemental folate intake but not for dietary folate only (P = .03 and P = .64 for linear trend, respectively). Greater frequency of multivitamin use showed a modest suggestive inverse association (current multivitamin use of ≥6 times per week vs nonuse multivariable relative risk, 0.84; 95% CI, 0.64-1.11; P = .06 for linear trend).
Conclusions and Relevance
Higher total folate intake was associated with a suggestive lower risk for EG/SEG, supporting a possible causal role of homocysteine in EG/SEG.
Exfoliation glaucoma (EG), associated with exfoliation syndrome (ES), is the most common secondary open-angle glaucoma.1 In EG, intraocular pressure (IOP) may become elevated due to narrowing and increased pigmentation in the filtration angle,2 protein leakage into the anterior chamber,3 and exfoliation material buildup in the trabecular meshwork.4 Established EG risk factors include older age,5,6 lysyl oxidase–like 1 variants (LOXL1),7 and northern latitude residence.7- 9 Effective strategies for EG primary prevention are lacking.
One possible EG risk factor that has received substantial research attention is homocysteine. Elevated homocysteine may enhance exfoliation material formation by contributing to vascular damage,10 oxidative stress,11,12 and extracellular matrix alterations.13 Indeed, homocysteine levels in plasma,14- 19 aqueous humor,20 and tears21 have been consistently elevated with ES/EG. Lowering homocysteine levels may be an attractive target for intervention because this can be achieved by increasing intakes of vitamin B6, vitamin B12, and (most important) folate.22 Several case-control studies18,21,23- 25 have reported lower plasma folate levels with ES, but differences in vitamin B6 or vitamin B12 levels have not been found.19 Major limitations of these studies are the small sample size and cross-sectional design. To date, no prospective studies have evaluated B vitamin intake and risk for EG. Therefore, we conducted a prospective study among 78 980 women and 41 221 men followed up for more than 20 years of the association between intake of folate, vitamin B6, and vitamin B12 and EG or suspected EG (EG/SEG) risk.
The human research committees of Brigham & Women’s Hospital, Massachusetts Eye and Ear, and Harvard School of Public Health approved this study. Informed consent was implied by receipt of completed questionnaires in both cohorts. In 1976, the Nurses’ Health Study (NHS) began when 121 700 US female registered nurses aged 30 to 55 years responded to a mailed questionnaire.26 In 1986, the Health Professionals Follow-up Study (HPFS) began with 51 529 male health professionals who responded to a similar questionnaire.27 In these studies, participants have been completing biennial questionnaires about their diet, lifestyle, and newly diagnosed diseases, such as glaucoma. The follow-up rate was high (>85%).
The study period was from June 1, 1980 (baseline) to May 31, 2010 (study end) in the NHS and from January 1, 1986 (baseline) to December 31, 2010 (study end) in the HPFS. A participant contributed person-time if she or he was 40 years or older (because glaucoma risk increases after age 40 years) and indicated an eye examination in the 2-year risk period (to minimize detection bias). Participants contributed person-time in approximately 2-year units based on biennial questionnaire responses from baseline until the earliest occurrence of glaucoma, cancer, death, loss to follow-up, or study end (May 31, 2010, in the NHS and December 31, 2010, in the HPFS).
Of the original cohort members, participants were excluded at baseline for the following reasons: (1) they did not complete the initial Semiquantitative Food Frequency Questionnaire (SFFQ); (2) the SFFQ dietary data were inadequate (adequate dietary information was defined as >50 of 61 items completed, yielding 500-3500 kcal/d, for women or >60 of 131 items completed, yielding 800-4200 kcal/d, for men); (3) they had prevalent cancer (excluding nonmelanoma skin cancer); (4) they had prevalent glaucoma or suspected glaucoma; (5) they were permanently lost to follow-up within 2 years after baseline; (6) they had no report of any eye examination during the follow-up period; and (7) they had prevalent cataract extraction, which makes EG/SEG diagnosis difficult. By 2010, a total of 78 980 women and 41 221 men had contributed person-time.
From participants self-reporting glaucoma, we obtained permission to retrieve medical information. Diagnosing eye care providers were sent glaucoma questionnaires about maximum IOP, optic nerve features, filtration apparatus status, and the presence of exfoliation material or other secondary causes of elevated IOP and were asked to send all visual field reports; alternatively, they were asked to send complete medical records. A glaucoma specialist (L.R.P.), masked to dietary data, evaluated the available medical information in a standardized manner to confirm diagnoses.
We defined EG/SEG based on documentation of at least 1 of the following in the eye with exfoliation material: (1) IOP exceeding 21 mm Hg, (2) cup-disc ratio of 0.6 or higher, or (3) visual field loss consistent with glaucoma on at least 1 reliable test. Patients with ES only who did not meet these criteria were excluded from analysis because we initially asked all participants about any glaucoma diagnoses, and our intent was to identify different types of glaucoma cases.
During the study period, 8029 women and 3422 men reported new glaucoma diagnoses. This was confirmed in 64.1% of women and 56.0% of men as follows: EG/SEG (4.8% of women and 2.7% of men), primary open-angle glaucoma (26.3% of women and 27.1% of men), elevated IOP or optic disc cupping only (18.8% of women and 16.4% of men), and other types of glaucoma or suspected glaucoma (14.2% of women and 9.8% of men). The others were unconfirmed (35.9% of women and 44.0% of men) because participants (7.2% of women and 14.3% of men) or eye care providers (5.1% of women and 4.4% of men) could not be contacted, participants did not give permission for medical record review (11.7% of women and 9.6% of men), participants indicated the initial report was erroneous (9.7% of women and 14.2% of men), or eye care providers refuted the glaucoma diagnosis (2.2% of women and 1.5% of men). Ultimately, we included 329 women and 70 men who met the criteria for incident EG/SEG.
We collected dietary intake data repeatedly using the SFFQ every 2 to 4 years from June 1, 1980, in the NHS and from January 1, 1986, in the HPFS.28,29 The 1980 SFFQ included 61 food and beverage items, the 1984 SFFQ included 116 items, and similar versions were used from 1986 onward in the NHS (126 items) and the HPFS (131 items). The SFFQ lists foods and beverages, each with a serving size, and asks about the average intake during the past year (with 9 responses for intake frequency, ranging from “never or less than once per month” to “6 or more times per day”). Questions were included on the brands of breakfast cereal used (because these foods are an important source of B vitamins), as well as on the brands and duration of use of multivitamins taken.
Nutrient intakes were computed by multiplying the consumption frequency of each food by the nutrient content of the portion specified. Food nutrient contents were obtained from the Harvard University Food Composition Database, derived mainly from US Department of Agriculture sources30 and others; these sources have been continually updated, with incorporation of changes in food folate content after grain folate fortification.31 All nutrient values were total energy adjusted with the residuals method.32 In validation studies29,33 of the SFFQ compared with detailed 1-week diet records, the correlation coefficients between the SFFQ estimates and the dietary record estimates were 0.77 for folate, 0.85 for vitamin B6, and 0.56 for vitamin B12. Furthermore, the correlation was 0.56 between folate intake and red cell folate level,34,35 and vitamin B6 intake predicted plasma pyridoxal phosphate levels (r = 0.52).
For each cohort, we calculated the cumulatively updated intakes for each B vitamin by averaging intakes from all available dietary assessments up to the start of each 2-year risk period. Because glaucoma develops slowly, we chose to study cumulatively averaged intakes because they best represent long-term diet; also, cumulatively averaged intake estimates have inherently less measurement error than estimates from single assessments.36
For age-adjusted and multivariable analyses that controlled for potential EG/SEG risk factors simultaneously, we used Cox proportional hazards regression analysis stratified by age in months and the specific 2-year risk period37 to estimate incidence relative risks (RRs) and their 95% CIs. We conducted tests for trend by including the median values within each intake category of quintiles (Qs).
We first analyzed each cohort’s data separately. Then, we pooled the results using meta-analytic methods incorporating random effects with testing for any heterogeneity across cohorts.38
Multivariable models were used in our primary analyses. These included the following covariates: cigarette smoking (in pack-years), cumulatively updated intakes of alcohol (in grams per day), total calories (in kilocalories per day) and caffeine (in milligrams per day), family history of glaucoma (glaucoma in biologic parents, siblings, or children), body mass index (calculated as weight in kilograms divided by height in meters squared), major ancestry (Scandinavian Caucasian, Southern-European Caucasian, other Caucasian, or other ancestry), and lifetime average continental US residence categorized by latitude (northern is >42° N, middle is 37°-42° N, and southern is <37° N), as well as self-report (yes or no) of hypertension, diabetes mellitus, high cholesterol level, and myocardial infarction.
For secondary analyses, we evaluated timing of exposure by investigating baseline intake only and the most recent intake only. Also, we separately evaluated intakes from different sources (ie, dietary only or supplemental only). This may be most pertinent for vitamin B12 and folate, which are more bioavailable in supplemental form. For example, supplemental folate in the form of folic acid is more bioavailable than natural folate (1 μg of food folate equals 1.0 μg of dietary folate equivalents, but 1 μg of folate supplement equals 2.0 μg of dietary folate equivalents).39,40
To investigate effect modification by other factors for folate, we examined associations separately by family history of glaucoma, lifetime residential history, and alcohol intake (because alcohol can interfere with folate metabolism).9,22 We evaluated for effect modifications by testing interaction terms in Cox proportional hazards regression models.
The source of supplemental B vitamins in this population was mostly from multivitamins. Therefore, we evaluated the association between multivitamin use and risk for EG/SEG.
During the follow-up period with 1 347 004 person-years in the NHS and 494 822 person-years in the HPFS, we identified 329 incident EG/SEG cases in the NHS and 70 cases in the HPFS. As expected, at diagnosis, those with EG/SEG were likely to have high IOP (28-29 mm Hg), have one affected eye (53%-64%), and be older than 65 years (mean [SD] age, 68  years in women and 71  years in men).
Participants having the highest B vitamin intake generally used less alcohol, caffeine, and cigarettes; were more likely to have lower body mass index; and were less likely to have lived in the northern tier states compared with participants having the lowest intake (Table 1). All these differences were accounted for in multivariable analyses.
Age-adjusted and multivariable analyses were similar, as were the results in men and women (Table 2). In pooled results, an inverse trend was found between higher total folate intake and risk for EG/SEG (P = .02 for linear trend). Compared with Q1 of total folate intake (median, 217 μg/d in women and 282 μg/d in men), the multivariable RR (MVRR) of EG/SEG was 0.75 (95% CI, 0.54-1.04; P = .02 for linear trend) for Q5 (median, 654 μg/d in women and 839 μg/d in men). These results were minimally changed when we also adjusted for vitamin B6 and vitamin B12 intake: the MVRR for Q5 vs Q1 was 0.75 (95% CI, 0.52-1.07; P = .03 for linear trend).
To evaluate folate intake by source, we simultaneously included dietary folate and supplemental folate in multivariable models. We observed no trends with dietary folate intake only (P = .64 for linear trend): compared with Q1 of dietary folate intake (197 μg/d in women and 259 μg/d in men), the MVRR for EG/SEG was 0.97 (95% CI, 0.70-1.35) for Q5 (377 μg/d in women and 496 μg/d in men) (Table 2). However, we observed inverse trends with greater supplemental folate intake (P = .03 for linear trend): compared with Q1 of supplementary folate intake (0 μg/d in women and men), the MVRR for EG/SEG was 0.83 (95% CI, 0.54-1.28) for Q5 (335 μg/d in women and 434 μg/d in men).
For timing of exposure, the association between total folate intake at baseline was null, while the association with the most recent diet was similar to that with the cumulatively averaged diet (main analysis). The MVRR for Q5 vs Q1 was 0.73 (95% CI, 0.53-1.02; P = .02 for linear trend) (Table 2).
We observed no interactions of folate intake with alcohol use, family history, or geographic location (P > .60 for interaction) (Table 3). Nonetheless, suggestive stronger inverse associations were observed for those with higher intake of alcohol (MVRR for Q5 vs Q1, 0.63; 95% CI, 0.33-1.21; P = .03 for linear trend), those without a family history (MVRR for Q5 vs Q1, 0.72; 95% CI, 0.49-1.08; P = .09 for linear trend), and those who have lived mostly in the northern tier states (MVRR for Q5 vs Q1, 0.73; 95% CI, 0.47-1.15; P = .02 for linear trend). We observed no interactions between supplemental and dietary intakes (P = .29 for interaction); however, inverse trends were observed for greater supplemental folate intake among those with dietary folate intake less than the median (MVRR for Q5 vs Q1, 0.65; 95% CI, 0.33-1.27; P = .01 for linear trend).
Overall, we observed no associations between higher total vitamin B6 intake and risk for EG/SEG (Table 4). Compared with Q1 (1 mg/d in women and 2 mg/d in men), the MVRR for EG/SEG was 0.86 (95% CI, 0.53-1.41) for Q5 (14 mg/d in women and men). The point estimate was closer to the null when we also adjusted for folate and vitamin B12 intake: the MVRR for Q5 vs Q1 was 1.05 (95% CI, 0.67-1.62; P = .53 for linear trend). Little variation was found in dietary vitamin B6 intake, and associations were null. For supplemental B6, the MVRR for Q5 vs Q1 was 0.77 (95% CI, 0.50-1.18; P = .45 for linear trend). Analysis results of baseline intake or the most recent intake of vitamin B6 were similar to the results of the main analysis.
Similarly, we observed no trend (P = .99 for linear trend) between higher total vitamin B12 intake and risk for EG/SEG (Table 5). Compared with Q1 (5 μg/d in women and 6 μg/d in men), the MVRR for EG/SEG was 0.88 (95% CI, 0.64-1.21) for Q5 (16 μg/d in women and 22 μg/d in men). The point estimate was closer to the null when we also adjusted for folate and vitamin B6 intake: the MVRR for Q5 vs Q1 was 1.11 (95% CI, 0.75-1.63; P = .25 for linear trend). The association with dietary vitamin B12 intake was null. For supplemental B12 intake, the MVRR for Q5 vs Q1 was 0.59 (95% CI, 0.36-0.98; P = .47 for linear trend). The point estimate was closer to the null with adjustment for folate and vitamin B6 intake: the MVRR for Q5 vs Q1 was 0.68 (95% CI, 0.39-1.18; P = .81 for linear trend). Analysis results of baseline intake or the most recent intake of vitamin B12 were similar to the results of the main analysis.
Greater frequency of multivitamin use showed a modest suggestive inverse association. For current multivitamin use of at least 6 times per week vs nonuse, the MVRR was 0.84 (95% CI, 0.64-1.11; P = .06 for linear trend).
We observed a suggestive trend of lower risk for EG/SEG with higher total folate intake, supporting a possible causal role of homocysteine in EG/SEG. Inverse associations were stronger with supplemental folate intake (which is more bioavailable than natural folate40), especially among those with low dietary folate intake. For folate intake, no interactions were observed with alcohol use, family history, or residential latitude. No associations were observed with baseline intakes, indicating that consistent long-term and more recent intakes of B vitamins may be more causally relevant. We observed no associations with intake of vitamin B6 or vitamin B12. Consistent with a previous study41 showing associations between EG/SEG and higher coffee intake, which also increases homocysteine,42,43 the inverse association with lower folate intake further implicates homocysteine in the origin of EG/SEG. Because this is the first prospective study to date to evaluate long-term B vitamin intake and risk for EG/SEG, our results must be interpreted cautiously.
The association of lower risk for EG/SES with higher folate intake only and not with vitamin B6 or vitamin B12 intake is consistent with a 2012 meta-analysis25 of case-control studies of plasma levels of homocysteine, folate, vitamin B6, and vitamin B12. Compared with the level in controls, the mean blood homocysteine level in cases was higher, and blood levels of folate (but not vitamin B12 and vitamin B6 levels) were lower in cases.21 While all these vitamins are likely important for lowering homocysteine, most participants had intakes of vitamin B6 and vitamin B12 that exceeded the recommended dietary allowance of approximately 1 mg and approximately 2.4 µg, respectively, while more than 50% of participants had low folate intake (recommended dietary allowance, 400 μg)40; therefore, we had more variability in folate intake, which allowed for detection of associations. From the diet, folate is commonly found in fruits and vegetables, and in Europe, where supplement use and folate fortification are less common than in the United States, the Reykjavik Eye Study44 showed that higher consumption of fruits and vegetables was protective against ES, consistent with our findings. Future studies should assess these possible associations with B vitamin intake, evaluate biomarkers in a prospective design, and investigate gene-environment interactions. Although the 2012 meta-analysis25 of genetic variants of homocysteine metabolism enzymes showed no overall associations, the investigations have been few, and none have evaluated gene-environment interactions.
Our study’s strengths include its prospective design, whereby diet was assessed before disease onset, and the fact that it is a long-term large study with high follow-up rates. Other strengths include repeated updating of diet and lifestyle risk factors, which allowed us to examine intake at various periods, and the evaluation of nutrients from different sources. Finally, with the wealth of follow-up data, we had information on major potential confounders, which were adjusted for to minimize confounding bias.
Major limitations include the fact that it was not feasible to conduct in-person eye examinations and that EG/SEG cases were identified by self-report and then confirmed by medical records. However, using this approach, it was previously confirmed that incident EG/SEG is a strongly age-related condition that produces higher IOPs at diagnosis than incident primary open-angle glaucoma9 and that northern latitude residence is a risk factor.9 This indicates our ability to validly detect established risk factors for EG/SEG.6 Our study could not identify individuals with only ES without any signs of glaucoma, and we likely had substantial underascertainment of EG/SEG. However, our intent was not to ascertain absolute incidence rates but to evaluate the relative rates by categories of potential risk factors. For such a goal, it has been methodologically established that a low sensitivity for disease identification does not cause biases if the disease definition has high specificity and the underascertainment is similar across exposure groups,45 which applies to our study. A limitation is that higher B vitamin intake may be a marker of another unmeasured risk factor, and we may have had some residual confounding. Finally, this population was predominantly of white race/ethnicity and well nourished, limiting the generalizability of results more broadly.
In this large prospective study, higher intake of folate was associated with lower risk for EG/SEG. This further supports the hypothesis that homocysteine may have a causal role in EG/SEG.
Submitted for Publication: August 13, 2013; final revision received October 25, 2013; accepted November 24, 2013.
Corresponding Author: Jae H. Kang, ScD, Channing Division of Network Medicine, Department of Medicine, Brigham & Women’s Hospital, and Harvard Medical School, 181 Longwood Ave, Boston, MA 02115.
Published Online: April 3, 2014. doi:10.1001/jamaophthalmol.2014.100.
Author Contributions: Dr Kang had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Kang, Willett, Pasquale.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Kang, Wiggs, Willett, Pasquale.
Critical revision of the manuscript for important intellectual content: Kang, Loomis, Willett, Pasquale.
Statistical analysis: Kang, Loomis, Willett.
Obtained funding: Wiggs, Willett.
Administrative, technical, or material support: Kang, Wiggs, Pasquale.
Study supervision: Pasquale.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by grants P01 CA87969, UM1 CA167552, EY09611, and EY020928 (Dr Wiggs) and grant EY015473 (Dr Pasquale) from the National Institutes of Health and the Arthur Ashley Foundation. Drs Wiggs and Pasquale are also supported by the Harvard Glaucoma Center of Excellence.
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.