| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
1 Departments of Pediatrics and Environmental Health Sciences, Columbia University; 2 Ophthalmic Oncology Service, Department of Surgery, Pediatrics, Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, New York; 3 Departments of Ophthalmology and Social Work, Instituto Nacional de Pediatria; 4 Department of Ophthalmology, Hospital Infantil de Mexico; 5 Hospital Siglo XXI, Instituto Mexicano de Seguro Social, Mexico City, Distrito Federal, Mexico; and 6 Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts
Requests for reprints: Manuela Orjuela, Department of Environmental Health Sciences, Mailman School of Public Health at Columbia University, Room B106, 60 Haven Avenue, New York, NY 10032. Phone: 212-304-7279; Fax: 212-544-1943. E-mail: mao5{at}columbia.edu
 |
Abstract |
|---|
 Top Abstract BackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
Methods: At the Instituto Nacional de Pediatria in Mexico City, we conducted a hospital-based case-control study to evaluate prenatal maternal diet. We examined dietary intake of fruits and vegetables of mothers of 101 children with retinoblastoma and 172 control children using a dietary recall questionnaire and published food nutrient content tables.
Results: The reported number of mean daily servings of fruits and vegetables was lower among case mothers when compared with control mothers [vegetables: 2.28 in controls, 1.75 in cases (P < 0.01); fruits: 2.13 in controls, 1.59 in cases (P = 0.07)]. Mean daily maternal folate intake from both vegetables and fruits was higher in controls (103 µg) than in cases (48 µg; P < 0.05). Risk for having a child with retinoblastoma was increased for mothers consuming fewer than 2 daily servings of vegetables [odds ratios (OR), 3.4; 95% confidence interval (95% CI), 2.0-6.0] or with a low intake of folate (OR, 3.9; 95% CI, 2.1, 7.3), or lutein/zeaxanthin (OR, 2.6; 95% CI, 1.5-4.6) derived from fruits and vegetables.
Conclusions: Decreased intake of vegetables and fruits during pregnancy and the consequent decreased intake of nutrients such as folate and lutein/zeaxanthin, necessary for DNA methylation, synthesis, and retinal function, may increase risk for having a child with sporadic retinoblastoma.
|
Background |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
Retinoblastoma arises as the result of mutations in both alleles of the retinoblastoma gene, RB1. These mutations occur prezygotically in germ cells (initial mutation in bilateral disease) or postzygotically (mutation in the second allele in bilateral disease; mutations in both alleles in unilateral disease) in retinal cells during fetal development or early infancy (2). Although much is understood about the effects of the RB1 mutations underlying the formation of retinoblastoma, little is known about the etiology of these mutations. New germ-line mutations are known to occur preferentially on the paternal allele, suggesting the implication of paternal preconceptional risk factors (4). However, there is no knowledge about the etiology of retinal cell mutations or their time of occurrence. In retinoblastoma, the crucial period for development of mutations in retinal cells is likely to be during retinal formation, which begins between the 4th and 8th week of gestation (5), and continues through early infancy, with macular differentiation occurring at about 6 months (6).
The incidence of unilateral retinoblastoma has been reported to be higher in several less affluent regions of the world (7) suggesting that environmental factors associated with poverty may increase the risk of mutagenesis in retinal cells. Incidence of retinoblastoma in Mexico City is 15.1 cases per million infants, and 12.2 per million children between the ages of 1 and 4 (8). Among children in the United States, incidence per million children under 5 years is 12 in whites and 15 in blacks (9, 10). A study done within the Mexican Institute of Social Security found that incidence of retinoblastoma occurs primarily among children of blue-collar workers and seems higher in Chiapas, the poorest state of Mexico (11). In the United States, a low level of maternal education seems to confer an elevated risk for the development of sporadic retinoblastoma, whereas supplementation with multivitamins seems to be protective (12).
We hypothesize that decreased maternal micronutrient intake during retinal formation may contribute to development of sporadic retinoblastoma. In this study, we investigated whether such a decreased intake during pregnancy is associated with development of sporadic retinoblastoma.
|
Materials and Methods |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
Among cases, the diagnosis of retinoblastoma was made by the Instituto Nacional de Pediatria ophthalmologist treating the child, and then was verified histologically by the Instituto Nacional de Pediatria pathologist. Hospital charts of all participants were reviewed by one of the study investigators (M.O.) to verify eligibility criteria.
The interview included questions about household, occupational, and personal exposure during the pregnancy, mother's pregestational and gestational medical history, father's medical and occupational history, and medical history of the index child. For diet during the index pregnancy, mothers were asked three open-ended questions requesting that they list foods they had typically consumed for breakfast, lunch, and dinner, and for each food, the number of times per week consumed. Intake was calculated using the number of daily servings consumed for each food item assuming standard portion sizes, using the Tabla de Alimentos Mexicanos (13) and the Nutrition Data System for Research software (version 4.04, Nutrition Coordinating Center, University of Minnesota, Food and Nutrient Database 32; 2001). Mothers were asked whether they had had morning sickness which had impeded them from eating their regular amounts of food during the first trimester, as well as about their use of vitamin supplements including iron, folate, vitamin C, calcium, and multivitamins during the pregnancy.
Statistical Analysis
We compared the dietary intake of mothers of cases and controls with regard to all foods and groups listed by the mothers. Because we found that only fruit and vegetable intake differed between the two groups of mothers (see Results), we focused our subsequent analyses on fruits and vegetables. The numbers of servings of fruits and vegetables consumed daily were calculated using the Mexican nutrient table categorization of fruits and vegetables (13). Folate, B6, lutein/zeaxanthin (subsequently referred to as "lutein"), and 
- and ß-carotene are contained in similar foods. To examine nutrient consumption based on the daily servings of fruits and vegetables, a composite variable was created from the sum of folate, B6, or carotenoids ingested from all of the fruits and vegetables reported consumed. Each fruit or vegetable was then weighed by its nutrient content in microgram per 100 gram serving (13). Nutrient content was then multiplied by the number of daily servings reported consumed. To examine risk from low vegetable or fruit intake, vegetable and fruit intakes were first examined as categorical variables with the number of daily servings (range, 0-6). Low fruit and low vegetable intakes were then dichotomized, using consumption of 2 daily servings of either fruits or vegetables as a cut point. Bean intake was examined by dichotomizing intake, using the median daily servings in the controls (1.0 daily serving) as the cut point.
Risk from intake of low levels of nutrients (folate, -carotene, ß-carotene, lutein, and vitamin B6) derived from fruits and vegetables was examined by dichotomizing the variable using the median value for intake found in the control population (84 µg for folate, 149 µg for -carotene, 8,021 µg for ß-carotene, 650 µg for lutein, and 370 µg for B6) as the cut point. To examine the association between intake and outcome for lycopene and ß-cryptoxanthine, we examined presence versus absence of nutrient intake.
Descriptive statistics were computed for mothers of cases and mothers of controls (Table 1). Categorical and continuous variables were examined using 
2 and Wilcoxon's rank sum tests to test for differences between the two groups of mothers. The Pearson correlation coefficient was used to examine the relationship between continuous variables. Logistic regression was used to generate unadjusted and adjusted odds ratio (OR) controlling for potential confounding variables. All analyses were done comparing controls with total cases (bilateral, unilateral, and unknown laterality cases combined), as well as with unilateral cases alone. Cases included in the unilateral disease analyses were those for which diagnosis of unilateral disease was confirmed (by documenting that disease had not developed in the second eye). Many cases were lost to follow-up before definitive confirmation of their laterality. Potential confounders we considered in our final model included maternal education, maternal age, place of residence, prenatal maternal vitamin intake, household and per capita income, and maternal alcohol use. Household and per capita income were examined as dichotomous variables using the median income in controls as the cut point. Maternal education was dichotomized by completion of 9 years of education (secondary school in Mexico ends at 9th grade). Place of residence was examined using a dichotomous variable for living inside or outside of the metropolitan Mexico City region. Variables were included in the final model if they were significant predictors of outcome in univariate analyses and if their inclusion in the final model modified the point estimates of the effects of other predictors by at least 10%.
|
|
Results |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
A similar proportion of case and control mothers reported taking some form of vitamin supplement at any point during their pregnancies. We did not ask mothers about the frequency or duration of their supplement use nor about the timing of initiation of supplementation with individual vitamins. However, we did ask when they first took "any form of vitamin supplement."
There was a slightly higher proportion of boys among cases (63.5%) than controls (57.5%). Because there is seasonal variation in the availability of fruits and vegetables in central Mexico, we examined whether there was a difference in the months during which the pregnancies of cases and controls transpired by comparing months of birth for the two groups. Births were evenly distributed throughout the year in both cases and controls with a slightly higher proportion of births occurring in January, September, and October in both groups (data not shown). Overall, there was no difference in the distribution of births throughout the year between the two groups (P = 0.49).
Appendix 1 shows the reported number of daily servings of all foods reported consumed by the mothers participating in our study. Consumption was similar between case and control mothers for most foods. However, cases ate significantly fewer servings of oatmeal, chicken, watermelon, cantaloupe, pears, carrots, peas, chayote, string beans, zucchini, beets, broccoli, and spinach. Consumption of fresh cheese, beans, and tortillas were higher among cases, although the differences were only of borderline significance. Because consumption differed primarily in fruit and vegetable intake, we focused our subsequent analyses on these two food groups. The reported number of daily servings of fruit and vegetable sources was significantly lower among case mothers (Table 2). The average number of daily servings of fruits was 1.59 in cases and 2.13 in controls (P = 0.07); that for vegetables was 1.75 in cases and 2.28 in controls (P < 0.01). Higher daily consumption of each of the following foods was associated with reduced risk of retinoblastoma in the univariate analysis: watermelon, cantaloupe, peaches, pears, chayote, carrots, green peas, string beans, squash, beets, broccoli, and spinach (data not shown).
|
- and ß-carotene, B6, and folate derived from fruits and vegetables between case and control mothers. Average daily maternal intakes of these nutrients were significantly lower in cases than in controls (Table 2; Fig. 1). Folate, B6, lutein, and - and ß-carotene contents for fruits as well as vegetables were significantly correlated with each other. Dietary B12 consumption also did not differ between cases (5.2 µg) and controls (6.1 µg; P = 0.06).
|
|
|
|
In conclusion, low vegetable intake is associated with increased risk for development of sporadic retinoblastoma, even after adjustment for low maternal education and living outside the capital. Some nutrients contained in fruits and vegetables have an independent additional effect on this risk. Among these, -carotene, lutein, and folate are highly correlated. Therefore, it is difficult to separate out their individual effects. Despite this, low intake of folate derived from fruits and vegetables seems to be the more important individual nutrient risk factor for having a child with sporadic retinoblastoma.
|
Discussion |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
Lower intake of fruits and vegetables during pregnancy was associated with increased risk for having a child with retinoblastoma. Intake of fruits and vegetables may be protective against development of malignancy. Adults who consume diets rich in these foods have lower levels of DNA adducts in their leukocytes, suggesting lesser DNA damage or more efficient repair of that which occurs (16). Fruit and vegetable intake during pregnancy has been shown to be protective against development of medulloblastoma, another primitive neuroectodermal tumor (17). In our population, the effect of decreased intake of some micronutrients resulting from the lower intake of fruits and vegetables was stronger than that resulting from decreased intake of the foods themselves. Of the nutrients examined, folate seemed to have the strongest effect, although other nutrients such as B6 and lutein were also important contributors. The effect of B6, although not independent of that of folate, is noteworthy as B6 (pyridoxine) has been shown to promote DNA excision repair, decreasing levels of unrepaired damaged DNA (18). Higher plasma levels of B6 are associated with a decreased risk for development of lung (19) and breast cancer (20). Similarly, the effect of lutein was not independent of that of folate in our study sample, although both nutrients are contained in some of the same foods and their intake is highly correlated in our population. The antioxidant lutein (and zeaxanthin, its isomer) is selectively absorbed by the retina and concentrated in the macula, forming the macular pigment (21-23). Supplementation with lutein/zeaxanthin has been associated with improvement in visual function in patients with age-related macular disease (24, 25). Lutein may be important during retinal formation, and its antioxidant properties may deter formation of retinal mutations. It is possible that it is a combined effect of B6, lutein/zeaxanthin, and folate that contributes to the observed protection derived from fruits and vegetables in our study population.
Folate is normally found in high concentrations in dividing neural cells and is required for synthesis of nucleotides (26). Deficiency of folate can lead to mutagenesis through impaired methyl group transfer and pyrimidine synthesis and through the misincorporation of uracil into DNA (by impairing conversion of dUMP to dTMP) leading to chromosomal breaks (27). Folate depletion is associated with increased risks of cervical (28), colon (29), and esophageal cancer (30). Maternal deficiency of folate during early gestation is associated with the increased incidence of neural tube defects (31, 32) and congenital cardiac anomalies (33). Maternal ingestion of pharmaceutical folate antagonists during pregnancy also increases the risk of cardiovascular, palatal, and urinary tract defects (34). Low folate levels may increase the risk of tumors in tissues that arise from the same embryonic layer as the neural tube (17).
Methylation, an epigenetic modification of DNA, is involved in the suppression of gene expression and enhancement of genomic stability (35). Hypomethylation can result from prolonged depletion of folate in humans (36) and, when present within gene-coding regions, has been associated with genomic instability leading to deletions, such as those found in RB1 in retinoblastoma (37).
Although naturally occurring folate is found in leafy green vegetables, some fruits, egg yolks, and legumes, the bioavailability of dietary folates obtained from natural sources is highly variable (2.9-72.2%), and depends on gastric acidity, food preparation, and exposure to air oxidation (38). For legumes, the bioavailability is particularly low; for example, only 4.5% of the folate in lima beans is bioavailable after cooking them for 10 minutes. Although folic acid in fortified foods has greater bioavailability, only 30% to 40% of folic acid in fortified bread is bioavailable (39). In Mexico, although some foods such as commercially available cereals and sliced bread are fortified with synthetic folic acid, most poor women, especially those living in more rural areas, do not have access to these fortified foods (40). Traditional diets in Central Mexico rely on corn flour and freshly baked bread. Beans are cooked for long periods of time, decreasing the bioavailability of naturally occurring folates. Folate deficiency is not uncommon in Mexico, particularly among women of childbearing age. In Northern Mexico, 44% of women of low socioeconomic status and 68% of women of middle socioeconomic status have been found to consume less than the recommended daily allowance for folate in women of childbearing age (400 µg); whereas 33% of middle socioeconomic status women studied had low serum and RBC folate levels (41).
Poor women, particularly those in rural areas, may have few options for modifying their diet during pregnancy. For our population, fruits and vegetables represented the major sources of bioavailable dietary folate. Although the relatively high use of multivitamin supplements would suggest that women were obtaining some folate from supplements, we do not have data on the frequency or duration of this intake. Subsequent work by our group suggests that vitamin intake is highly variable among populations similar to ours.
Other investigators who have examined a US population with retinoblastoma found a protective effect of multivitamin supplements during pregnancy (12). In our population, the observed protective effects do not seem to be due to multivitamin supplementation. There was no difference in reported use between case and control mothers, although fewer case mothers reported consuming multivitamins when compared with controls. Early gestational multivitamin consumption in our population seems similar to that of other poor populations in Mexico (41).
The major limitations in our study include possible recall and selection biases inherent in a case-control study and the crudeness of the dietary instrument that we used. We interviewed mothers up to 6 years after their pregnancies had taken place. One study in the United States has shown that mothers remember their diet during an index pregnancy with about 70% to 79% accuracy 3 to 7 years after the pregnancy has occurred (42). Another study of a US population evaluated maternal recall of diet during pregnancy and found that case and control mothers interviewed during pregnancy and then again 4.5 years later did not differ significantly in their recall of dietary intake during their pregnancy (43). However, no studies have specifically examined dietary recall in a case-control setting for children with malignancy in an impoverished population. Selection bias would take place if health-conscious parents with good diets were more likely to participate as controls. However, both case and control mothers came seeking treatment for ill children of the same age group in the same tertiary care public hospital, and participation refusal rates for control mothers were extremely low (0.17% for control mothers compared with 0.9% for case mothers). In addition, case and control mothers exhibited similar behavior in such activities as smoking (data not shown) and drinking, and came from comparable household income levels (Table 1).
Additional selection bias might have resulted from seasonal variation in availability of fruits and vegetables in central Mexico if we had found a difference for season of birth. The differences in the severity of the diagnoses of case and control children might contribute to recall bias. Compared with mothers of children with other nononcologic diagnoses, mothers of children with retinoblastoma might be expected to think harder about their diets during their pregnancies and thus potentially overreport or more correctly report their ingestion of fruits and vegetables. It is unlikely that they would have selectively underreported their intake. Such differences in recall would underestimate the magnitude of the association (44).
The major limitation in this study is the collection of dietary information through open-ended questions rather than through the use of a validated food frequency questionnaire. However, other investigators working in Mexico have used open-ended questions regarding fruit and vegetable consumption as part of food frequency questionnaires and have been able to correlate them with 5-day food registries (Pearson correlation = 0.7, P = 0.0001), which in turn correlated with serum folate levels (Pearson correlation = 0.4, P = 0.02; ref. 41).
Although our interviewer was not blinded to the case-control status of participants, and thus represented a potential source of observer bias, she was blinded to the study hypotheses, and would therefore have been unlikely to prompt dietary responses differentially between the two groups. Although information bias might result because we inquired about an average diet during pregnancy, and diet may change dramatically throughout the pregnancy, work of other investigators suggests that Mexican women in the socioeconomic group of our mothers are unable to substantially modify their dietary intake of micronutrients during their pregnancy (41, 42).
We did not collect information regarding serving size or manner of food preparation. However, we were able to make generalizations regarding food preparation given the work done by others who have studied populations similar to ours (45, 46). Vegetables are generally cooked and fruits are generally eaten raw in Mexico. We assumed that the portions consumed were of standard size. Preparation or portion size is unlikely to vary between case and control groups given their similar socioeconomic status.
The results of our study are consistent with those of other studies, which have found that other primitive neuroectodermal tumors occurring in early childhood are also related to maternal gestational micronutrient deficiencies. Investigators of the Children's Oncology Group found that maternal vitamin intake during pregnancy may significantly decrease the risk of having a child with neuroblastoma (47), whereas an earlier study found that maternal consumption of vegetables, fruits, multivitamins, and folate supplements during pregnancy significantly decreased the likelihood of having a child with medulloblastoma (17). Most recently, the incidence of neuroblastoma has been found to have decreased with the onset of folate supplementation in Canada (48). Together, these studies suggest the potential role of micronutrients in the development of early childhood malignancy.
Although our study was carried out in a population without access to folate fortified foods, we believe that our results are relevant to other populations, including those with fortified foods. Data from the third National Health and Nutrition Examination Survey (1988-1994) showed that more than 90% of US women of childbearing age who were surveyed had inadequate intakes of folate from food sources alone (49, 50). Since 1998, fortification of enriched grains with 140 µg folic acid/100 g is mandatory in the United States (51). However, researchers who examined the effect of current fortification methods in a group of educated, motivated women showed that with current practices, 61% of women of childbearing age had intakes of synthetic folic acid below the recommended level of 400 µg/d, and only those using supplements containing folic acid met the recommended intakes (52). This deficiency may be greater among impoverished or less educated populations.
Our data are consistent with our hypothesis that maternal nutritional exposures during gestation may influence the risk for a child developing retinoblastoma. Maternal nutritional exposures during gestation may mediate the effect of other environmental carcinogens on the developing fetus. Decreased intake of vegetables and fruits during pregnancy and the consequent decreased intake of nutrients such as folate and lutein, necessary for neural tissue development and retinal function, may be associated with an increased risk for having a child who develops sporadic retinoblastoma. Children exposed to low levels of maternal folate during retinogenesis may be more likely to develop mutations in their retinal cells, thus potentially influencing development of post zygotic mutations in RB1. Folate deficiency during retinogenesis could lead to increased uracil misincorporation or hypomethylation with increased mutagenesis in dividing retinal cells. This risk may be mediated by maternal genetic variation in genes regulating folate metabolism. The next useful step in testing this hypothesis includes examining such genetic variation, nutrient plasma levels, and dietary intake with validated food frequency questionnaires. Such research may provide additional insight into the development of this disease.
|
Appendix 1. Foods Reported Consumed by Mothers during Pregnancy |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
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.
Note: M. Ramirez-Ortiz, currently at the Hospital Infantil de Mexico.
Received 6/11/04; revised 3/28/05; accepted 4/ 8/05.
|
References |
|---|
|
TopAbstractBackgroundMaterials and MethodsResultsDiscussionAppendix 1. Foods Reported...References |
|---|
T polymorphism affect cancer risk: intake recommendations. J Nutr 2003;133:3748–53S.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
