| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
Clinical Nutrition Program, University of New Mexico, School of Medicine, Albuquerque, New Mexico
Address reprint requests to: Philip J. Garry, Ph.D., Clinical Nutrition Program, University of New Mexico School of Medicine, Rm. 215 Surge Building, 2701 Frontier Pl. N.E., Albuquerque, NM 87131
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Method: We conducted a 10 year longitudinal study (48 men and 77 women) and a one year cross-sectional study (165 men and 226 women) in healthy elderly men and women enrolled in the New Mexico Aging Process Study. Iron stores were estimated by serum ferritin concentrations and iron intake was determined by three-day food records in the longitudinal study and by a food frequency questionnaire in the cross-sectional study.
Results: We found no association between heme iron intake and iron stores in either the longitudinal or cross sectional study. In the cross-sectional study we found in women, but not in men, that age and supplemental iron intake were significantly and positively associated with increased iron stores.
Conclusion: Iron stores in elderly men are thought to reach steady state levels where iron absorption is adjusted to a level just sufficient to cover basal iron losses. In elderly women, we speculate that not enough time has elapsed for postmenopausal women to reach steady state levels of iron stores resulting in increases in iron absorption with age. Another factor is that use of hormone replacement therapy could further delay some women in reaching steady state iron levels due to continued menstrual blood losses.
Key words: iron intake, ferritin, iron stores, elderly men and women
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The association between increased body iron stores and disturbed glucose homeostasis was found in a cross-sectional study conducted in 1,013 middle-aged Finnish men after adjusting for age, family history of diabetes, body mass index (BMI), waist hip ratio (WHR), alcohol consumption, physical activity, use of oral diabetes medications and use of diuretics [1]. The Finnish study was based on the hypothesis that, while iron accumulation has to be severe to induce organ damage that leads to diabetes, a lesser accumulation of iron can alter glucose and insulin homeostasis of the body.
Fernandez-Real et al. [2], in a cross-sectional study of NIDDM individuals, found that insulin resistance was independently related to serum ferritin levels in individuals without evidence of iron overload. However, this cross-sectional study did not enable Fernandez-Real et al. [2] to infer causality between serum ferritin and insulin resistance.
Elderly individuals, especially men, found to be heterozygous for the relatively common Cys282Tyr (G845A) and the His187Asp (H63D) mutations of the human hemochromatosis associated gene (HFE) have been shown to have moderately increased iron stores [5]. Since increased iron stores have been noted to be an independent risk factor for acute myocardial infarction [6,7], recent studies have examined whether individuals heterozygous for the Cys282Tyr mutation are indeed at increased risk for cardiovascular disease. Two European studies reported evidence that both men and women who are carriers of the HFE Cys282Tyr mutation have a two- to three-fold increased risk of cardiovascular disease [3,4].
With new evidence that even moderate increases in iron stores may increase the risk for developing certain age related chronic diseases, there has been renewed interest in re-examining the association between dietary iron intake and iron stores. Two recent studies addressed the issue of the effects of age, non-heme iron, heme iron and supplemental iron intake on iron stores in humans. The first study was that of Hallberg et al. [8], who used a direct method for examining iron absorption and its association with iron stores. These authors conducted a clinical trial in which iron absorption from the whole diet was measured for five days in 31 healthy men with a mean age of 29.4±10.8 years. Non-heme iron was labeled with an extrinsic iron tracer and heme iron was labeled similarly using hemoglobin biosynthetically labeled with a different radioiron tracer. Iron absorption was examined in relation to the amount of stored iron as estimated by serum ferritin levels. They found an inverse relation between total iron absorption and concentration of serum ferritin up to 60 µg/L. In subjects with serum ferritin levels >60 µg/L, Hallberg et al. found that iron absorption decreased to a level just sufficient to cover basal iron losses, implying that, at a serum ferritin concentration >60 µg/L, no further accumulation of iron stores will occur by dietary iron absorption and/or supplementation. Hallberg et al. also compared the absorption from non-heme and heme iron and concluded that the control of heme iron absorption is also very effective. Thus, the Hallberg et al. findings do not support a hypothesis that a high meat intake (implying a high heme iron intake) might lead to an accumulation of iron stores in normal men by bypassing the strong regulation of iron absorption.
The second study by Fleming et al. [9] in 1998 reported results that were opposite of that reported by Hallberg et al. [8]. Fleming et al. [9] investigated the consumption of various dietary components and iron stores in an elderly sample of the Framingham Heart Study participants. Serum ferritin was used as a measure of body iron stores in 634 free-living elderly men and women (67 to 93 years of age). Dietary intake during the previous year was assessed by a food-frequency questionnaire. After controlling for gender, age, body mass index, total energy intake, smoking and use of aspirin and other medications known to affect blood loss, they found four dietary factors that were significant associated with iron stores. Heme iron, supplemental iron, dietary vitamin C and alcohol were positively associated with serum ferritin. Fleming et al. [9] concluded that dietary factors, namely heme iron intake and iron supplementation, can influence the accumulation of iron stores in elderly subjects.
Based on the report of Fleming et al. [9], we decided to examine the effects of iron intake on iron stores in elderly men and women enrolled in the New Mexico Aging Process Study (NMAPS) from a longitudinal and cross-sectional perspective. In the longitudinal study, we used three-day food records to assess iron intake, while in the cross-sectional study, we used a food frequency questionnaire to assess usual iron intake. Iron stores were estimated using serum ferritin concentrations.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
50 mL of blood was obtained from each person between 0800 and 0930 hours for various immunologic and biochemical measurements. The volunteers were seen on a yearly basis at approximately the same time each year. The entire population was seen between January and November at a rate of six individuals per week. For the longitudinal analysis we used three-day food records and laboratory measurements of iron status (CBC and ferritin) collected yearly from 1984 to 1993. Subjects selected for this study had assays in 1984 or 1985 and in 1993 and fewer than five missing years of dietary records. There were 125 subjects (77 women and 48 men) that met these criteria. For the cross-sectional analysis, we used laboratory results (CBC and ferritin) from 1993 and 1996 in which a total of 165 men and 276 women were administered a food frequency questionnaire (FFQ) (one FFQ per subject). The Human Research Review Committee of the University of New Mexico School of Medicine approved this study. Informed consent was obtained from each participant.
Laboratory Measurements
Complete blood count measurements were conducted using a Coulter "S" instrument (Coulter Electronics, Hialeah FL). Serum ferritin determinations were performed using a two-site enzyme immunoassay (Ferrizyme Kit, Abbott Laboratories, Chicago, IL).
Dietary Intake Measurements for the Longitudinal Analyses
A research nutritionist instructed the volunteers on how to keep an accurate three-day food record. Each volunteer was asked to measure all food eaten for three consecutive weekdays, Monday through Friday, and record the intake on standard coding forms. Brand names, methods of food preparation and recipes for any mixed dish eaten during the period were also recorded. Commercial plastic food models were used as instructional aids to assist the volunteers in judging portion sizes. A diet scale and an instruction booklet designed for this study were also provided. This booklet stressed the need for accuracy, completeness and recording food items prepared from a recipe.
At the end of the three-day recording period, a research nutritionist visited the volunteers’ homes to review the diet records. At this time, each record was subjectively evaluated for completeness and accuracy, and the participants were asked to provide additional information about any unclear food item. If the subject used vitamin or mineral supplements, the brand name, contents and amounts of each nutrient were recorded for inclusion in determining total intakes. All food records were coded by food item and amount and then analyzed for nutrient composition. Data for 1984 through 1989 were analyzed using the Highland View Hospital-Case Western Reserve University (CWRU) Nutrient Data Base, revision 6 (1982 ed). (Case Western Reserve University, Cleveland, Ohio). Data for 1990 were analyzed by the CWRU Nutrient Data Base release 10, 1989. Data for 1991 through 1993 were analyzed using the Food Intake Analysis System (FIAS), version 2.0 (University of Texas at Houston). When we compared 31, three-day food records, randomly selected from 250 records collected in 1991, using the CWRU and FIAS, the calculated group mean dietary iron intakes were not significantly different by paired t test, 14.4±6.8 and 14.9±6.7 mg/day, respectively, r=0.91.
In addition, dietary data for 1984, 1987 and 1989 were used to analyze food groups. For this analysis, the records for these years were reanalyzed using CWRU data base release 10, 1989. Food codes for meat, poultry and fish consumption were identified in existing three-day diet records as previously described [11,12]. As noted in an earlier publication [11], less than 2% of dietary energy came from foods coded as mixed dishes containing meat, poultry or fish. This 2% included the vegetable, grain, dairy and other components of the mixed dishes as well as the meat, poultry and fish. Therefore, meat, poultry and fish intakes were analyzed with mixed dishes excluded, giving only a slight underestimate of meat, poultry and fish intake. Iron from meat, poultry and fish was used to estimate heme iron intake by assuming that 40% of meat, poultry and fish iron is heme iron [13].
Dietary Intake Measurements for the Cross-Sectional Analyses
We used an adaptation of the Health Habits and History Questionnaire (HHHQ) [14–16] that was administered to each participant in the NMAPS by a trained research nutritionist. The HHHQ was modified to include southwestern regional foods. As in the longitudinal study, iron from meat, poultry and fish was used to estimate heme iron intake by assuming that 40% of meat, poultry and fish iron is heme iron [13].
Estimates of Body Iron
For the longitudinal study, we estimated iron stores by the method of Cook, et al. [17]. Different levels of iron status require separate calculations to estimate body iron as described in detail in our recent publication [18]. For subjects with serum ferritin concentrations above 12 µg/L, the method reduces to the following equation:
 |
where 400 is the proportionality constant, ln is the natural logarithm and SF is serum ferritin in µg/L. All of the subjects in the longitudinal study had serum ferritin concentrations above 12 µg/L. We chose to estimate iron stores in milligrams in the longitudinal study so that direct comparisons could be made with other previous studies using the formula developed by Cook et al. [17]. For the cross-sectional analyses, we used serum ferritin concentrations per se to estimate iron stores so that direct comparisons could be made with previous results reported by Fleming et al. [9].
Statistical Methods
The longitudinal data set consisted of up to nine yearly measurements of iron intake and iron stores over the period 1984–1993 for each of 77 women and 48 men. Serum ferritin concentrations were not determined in 1991. A regression model was fit using the generalized estimating equation approach (GEE) of Liang and Zeger [19] to allow for the serial correlation between yearly measurements of iron stores. When using this method, it is necessary to specify a working correlation structure for the data. We used an exchangeable correlation structure, which assumes that the correlation is constant between any two yearly measurements of iron stores for an individual. This correlation is the same for all individuals. Observations between individuals are assumed to be independent. The distribution of iron stores is assumed to be normal. The regression model expressed iron stores as a linear function of gender, age, total caloric intake, body mass index and dietary and supplemental iron intake. Dietary iron intake was categorized into quartiles, and supplemental intake was categorized into three levels. We also fit a similar model using the data from years 1984, 1987 and 1989 where we could separate dietary iron into heme and non-heme components.
The GEE approach was useful for investigating the association between iron stores and iron intake at a single point in time. For each individual, iron stores and iron intake were regressed on year to provide slope parameters that could be interpreted as the average yearly changes in iron stores and iron intake over the 10 year period. Similar slopes were computed for total caloric intake and body mass index.
A normal linear regression model was fit that expressed the slope for iron stores as a linear function of gender, age in 1984, slope for total caloric intake, slope for body mass index and separate slopes for dietary and supplemental intakes of iron. The predicted value for 1988 from the individual linear regressions was used as a summary measure for each person that would represent his or her typical value for the measure over this time period. 1988 was chosen because it is near the midpoint of the time interval 1984–1993. This value was close to the observed value in 1988. The predicted value has the advantage that it is available for all persons, while the observed value for 1988 is missing for some individuals.
For the cross-sectional analysis, we employed a multiple linear regression model in which non-dietary (age, gender, BMI) and dietary factors (calories from alcohol, vitamin C, energy, total iron intake from diet (heme and non-heme iron) and supplements were examined in relationship with serum ferritin levels. In our cross-sectional analysis we attempted to analyze our data in a manner similar to that conducted in the study of Fleming et al. [9].
All analyses were performed using procedures of the SAS software [20]. The GEE regression was computed using a SAS macro provided by Dr. Zeger.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Iron stores were estimated from serum ferritin levels, and it might be argued that those with high iron stores had elevated ferritin levels due to the presence of inflammation, a condition known to cause an increase in serum ferritin levels. While we did not measure erythrocyte sedimentation rates (ESR) in our subjects between 1984 and 1993, a standard procedure to assess for inflammation, we did determine white blood cell counts (WBC) at yearly intervals in all of the subjects. The mean (SD) of the WBC for the men and women was 5.91±1.63 and 5.77±1.17x109/L, respectively, and none of the subjects had a WBC>11x109/L that would indicate acute infection [21]. In addition, the medical histories of all subjects were reviewed for indications of inflammatory diseases. One subject has a history of connective tissue disease, one subject has a history of unspecified arthropathy, two subjects have histories of cardiomyopathy and four have histories of cellulitis. No subjects have rheumatioid arthritis. However, many (55 women and 32 men) have histories of osteoarthritis. We added an indicator to our GEE models for presence of inflammatory disease, including osteoarthritis, but found no significant effect on iron stores. In a recent publication, in which we examined the impact of the mutations (HFE) associated with the development of hemochromatosis on iron stores in health elderly men and women in the NMAPS, we did have ESR data [5]. We found no association between ESR values and total iron stores, as estimated by serum ferritin concentrations. Thus, while we did not have ESR results for the present study, we feel that underlying inflammatory processes did not significantly influence ferritin levels in the longitudinal or cross-sectional study.
We used the generalized estimating equation method (GEE) to directly model the relationship between iron stores and iron intake. In the GEE model the observed level of iron stores in a given year is regressed on the observed iron intake for that same year. Data from all years are pooled. The GEE method takes into account the lack of independence of the iron stores from year to year. It can also handle the problem of missing data, although the method requires some assumptions on the form of the year-to-year correlation in iron stores. We specified that the correlation in iron stores be identical for any two years. In our data it seems likely that iron stores for two adjacent years will be more highly correlated than iron stores measured further apart in time. We have verified that this is indeed true. Correlations vary from about 0.67 for iron stores measured in 1984 and 1993 to about 0.92 for iron stores measured only one year apart. However, the GEE approach yields consistent estimators even if the correlation structure is misspecified, provided that the pattern of missing data is random [19].
The GEE regression model (Table 2) found predicted iron stores for men to be 916 mg, which would be equivalent to a geometric mean serum ferritin level of 118 µg/L. Iron stores in women were approximately 81 mg lower than for men (835 mg, equivalent to a geometric mean ferritin level of 97 µg/L), but this difference was not statistically significant. There was no relationship between age or total caloric intake and iron stores. The marginally significant relationship between iron stores and body mass index is interesting but unexplainable at this time. Iron stores were significantly higher by approximately 52 mg for persons in the upper quartile of dietary intake (>15.66 mg/day) when compared to persons in the lowest quartile (<10.15 mg/day). This finding suggests that iron intakes in this range, i.e., less than 10.15 mg/day, may be just sufficient to replace basal losses of iron but insufficient to increase storage iron. Table 2 also shows that those individuals consuming more than 18 mg of supplemental iron per day had iron stores that were 65 mg greater than those not consuming any iron supplement (p=0.011).
It is generally believed that the bioavailability of heme iron is greater than non-heme iron [13], and presumably increasing the amount of heme iron in the diet will result in more iron being absorbed. We were able to separate dietary iron intake into heme and non-heme iron in 3 years that dietary intake information was collected. Table 3 shows that there was no significant association between dietary iron intake and iron stores for either heme or non-heme iron intake. Again, we found a significant effect of supplemental iron intake on iron stores in this model (p=0.0428).
In our cross-sectional study we found a significant interaction effect between gender by total iron intake. Because of the interaction effect, we examined men and women separately (see Table 5). For men, the only significant effect on serum ferritin in men was BMI (estimate=0.0409, p=0.0444). At this time we do not have an explanation for this finding. For women, significant positive effects on serum ferritin levels were noted for age (p=0.0223) and supplemental iron intake (p=0.0030) but no effect of heme or non-heme iron intake.
To summarize the combined results in the present study, the following observations are noteworthy. First, our cross-sectional results are in agreement with our longitudinal findings in that we found no effect of heme iron intake on iron stores. Supplemental iron intake had a significant effect on iron stores in women, but not in men, in the cross-sectional study as well as having a significant positive effect on iron stores in our longitudinal analysis where women and men were analyzed together. Because of the relatively small sample size in the longitudinal analyses, we did not have enough power to examine women and men separately. The positive influence of iron supplementation in the longitudinal analyses may simply reflect the fact that approximately twice as many women than men were included in this study.
The finding that age and supplemental iron had a significant effect on serum ferritin levels in women, but not men, in our cross-sectional study can possibly be explained by the following results in which we examined the impact of HFE mutations on iron stores in elderly men and women in the NMAPS [5]. We reported that the carrier frequency of the HFE 845A mutation in our population was 12.2% and the carrier frequency of the 187G mutation was 19.2% for a single mutation; 2.4% were compound heterozygous, 845A/187G, and 2.4% were homozygous for the 187G mutation. More men, 15 of 28 (54%), with estimated iron stores in the upper quartile, 
1,050 mg, had a HFE mutation compared to 25 of 83 (30%) whose estimated iron stores were <1,050 mg (p<0.025). For women, 16 of 42 (38.1%) women with iron stores in the upper quartile, 961 mg, had a mutation and this value was not statistically different from 41 of 125 (32.8%) of women who had a mutation and had iron stores <961 mg. None of the following confounding variables, total iron intake, alcohol consumption and inflammatory processes, as determined by ESR, proved significant in explaining why men with a mutation have increased iron stores. From this study we concluded that having either a single chromosomal 845A and/or 187G mutation results in higher iron stores for men than if no HFE mutation were present. In other words, iron stores in these elderly men may have reached maximal levels as corresponding to our "setpoint" hypothesis for iron stores [22]. The setpoint theory states that healthy elderly individuals have different steady state levels of iron stores that are probably under genetic control. At steady state levels of iron stores, iron absorption is thought to be limited to amounts required to match that necessary to replace basal losses. The age at which an individual achieves his or her theoretical setpoint of iron stores has not been determined. We speculate that this may occur around age 40 in men and after menopause in women. It could be that not enough time has elapsed for our postmenopausal women to reach steady state levels of iron stores and therefore the significant increase in iron stores with age in women. Another factor to be considered is that approximately 25% of the women in the NMAPS are on hormone replacement therapy [23], and this could further delay reaching steady state iron stores due to continued loss of blood through menstruation.
We agree with Hallberg et al. [8] that there is probably a level of iron stores at which increased iron intake has little or no effect on increasing iron stores. However, we do not feel that the cutoff point at which this occurs corresponds to a serum ferritin concentration of 60 µg/L. Time, age and HFE mutations need to be considered. However, the present study results were in accordance with those of Hallberg et al. regarding absorption of heme- and non-heme iron. We also feel that there is an effective control of both heme- and nonheme-iron and increasing heme iron in the diet will not lead to an accumulation of iron stores in normal men by bypassing the strong regulation of iron absorption.
Differences between our cross-sectional findings and those of Fleming et al. [9], especially the effects of heme iron intake on iron stores, could be related to a number of factors including sample size differences, interaction effects, differences in dietary and biochemical methodology and measurement error in determining dietary iron intake. It is unclear which factor or factors noted above could definitively explain the differences noted between our findings and those of Fleming et al. However, we believe that women and men should be examined separately. Even though iron stores increase after menopause, the time period after menopause is critical in any analyses. In addition, the use of HRT, could also delay the achievement of maximum or steady state iron stores where increasing iron intake no longer affects the further accumulation of iron stores.
Based on our longitudinal and cross-sectional study findings, we have to conclude that increasing heme iron intake does not effectively increase iron stores in the elderly. Iron supplements, unless recommended by a physician to correct an anemic condition, may not be warranted for a healthy elderly individual consuming a balanced diet. This may be especially important for those individuals who are homozygous or heterozygous for the HFE mutations because of recent studies showing that even moderate increases in body iron may increase the risk for diabetes and cardiovascular disease.
|
FOOTNOTES |
|---|
Received October 1, 1999. Accepted December 1, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Marian and G. Sacks Micronutrients and Older Adults Nutr Clin Pract, April 1, 2009; 24(2): 179 - 195. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. L. Seaverson, J. S. Buell, D. J. Fleming, O. I. Bermudez, N. Potischman, R. J. Wood, L. Chasan-Taber, and K. L. Tucker Poor Iron Status Is More Prevalent in Hispanic Than in Non-Hispanic White Older Adults in Massachusetts J. Nutr., February 1, 2007; 137(2): 414 - 420. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. M. Blanck, M. E Cogswell, C. Gillespie, and M. Reyes Iron supplement use and iron status among US adults: results from the third National Health and Nutrition Examination Survey Am. J. Clinical Nutrition, November 1, 2005; 82(5): 1024 - 1031. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Chernoff Micronutrient requirements in older women Am. J. Clinical Nutrition, May 1, 2005; 81(5): 1240S - 1245S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. L. van der A, P. H.M. Peeters, D. E. Grobbee, J. J.M. Marx, and Y. T. van der Schouw Dietary haem iron and coronary heart disease in women Eur. Heart J., February 1, 2005; 26(3): 257 - 262. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-M. Liu, S. E Hankinson, M. J Stampfer, N. Rifai, W. C Willett, and J. Ma Body iron stores and their determinants in healthy postmenopausal US women Am. J. Clinical Nutrition, December 1, 2003; 78(6): 1160 - 1167. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Beard Dietary iron intakes and elevated iron stores in the elderly: is it time to abandon the set-point hypothesis of regulation of iron absorption? Am. J. Clinical Nutrition, December 1, 2002; 76(6): 1189 - 1190. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
