I05

Iron status and brain function: serum ferritinlevels associated with asymmetries of corticalelectrophysiology and cognitive perlormanc&’2Don M Tucker, PhD, Harold H Sandstead, MD, James G Penland, MS,

Steven L Dawson, PhD, and David B Milne, PhD

ABSTRACT Levels of serum femtin and iron were examined in relation to cognitiveperformance and quantitative EEG measures in 69 normal university students. Higher levels ofserum ferritin were associated with greater activation of the left hemisphere relative to the right,indicated by less power in the EEG spectra from left hemisphere electrodes. Iron status wassignificantly related to cognitive performance on two ofthe cognitive tasks, and these relationshipswere consistent with the EEG asymmetries: higher ferritin predicted greater verbal fluency butpoorer nonverbal auditory task performance. These results suggest that body iron stores arerelevant to specific neurophysiological processes supporting attention. Am J (Yin Nuir

l984;39:105-l 13.

KEY WORDS Iron nutrition, femtin, brain function, EEG, cognition, dopamine. human

Introduction

Iron-deficiency anemia was apparently arecognized illness in ancient Egypt and wasknown to Hippocrates ( 1). Chlorosis, namedthe “green sickness” because of the pallor itproduced, was found in the 17th century tobe responsive to iron therapy (1). Since itsrecognition in the medical literature, irondeficiency has been thought to be associatedwith psychological symptoms, but clear evi-dence has been elusive. Given the currentprevalence of iron deficiency (2), definitionof the relationship between iron nutritureand psychological function could be of ma-jor importance. Problems in characterizingthe neuropsychological effects of iron defi-ciency have been both theoretical and meth-odological. Without some knowledge of therole of iron in the brain, attempts to docu-ment psychological effects of iron deficiencymust resort to nonspecific and thus insensi-tive measures. Although recent studies haveprovided evidence that in children there aremeasurable cognitive effects of iron defi-ciency (2-4), the particular features of atten-tion or cognition that are impaired remainto be determined (5).

Research with iron deficiency in animals

has suggested that an impairment of dopa-minergic neurotransmission may underliethe frequently observed decrement in behav-ior activity (5-7). This hypothesis is consist-ent with the concentration of iron in braindopamine pathways (8) and may suggest theinvolvement of specific attentional controlsystems dependent on dopaminergic path-ways that mediate the cognitive impairmentin iron-deficient humans.

Current models ofthe neural systems con-trolling attention (9, 10) point to the impor-tance of the brain’s self-regulation of its ac-tivation level as a function of the demandsfor cognitive activity. The simultaneous de-termination of cortical activation and cog-nitive performance may allow a precision ofneuropsychological measurement not pos-

I From the Grand Forks Human Nutrition ResearchCenter, Agricultural Research Service, United StatesDepartment of Agriculture and Department of Psy-chology, University ofNorth Dakota, Grand Forks, ND58202.

2 Address reprint requests to: Don M Tucker, PhD.Research Psychologist, Grand Forks Human NutritionResearch Center, P0 Box 7 166, University Station,Grand Forks. ND 58202.

ReceivedApril II, 1983.Accepted for publication August 9. 1983.

The .1�nerican Journa/ of(/i,zica/ !‘�l1!ri1ion 39: JANUARY 1984. pp 105-1 13 Printed in U.S.A.(C 1984 American Society for Clinical Nutrition

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

106 TUCKER ET AL

sible with conventional psychometrics, andshould be directly relevant to the functionof the neurotransmitter systems modulatingneural activation. We have applied measuresof cortical electrophysiology and cognitiveperformance to examine the role of ironnutriture in attentional processes.

In an initial study (1 1), iron status wasfound to covary with a number of EEGvariables, although no significant relation-ships were found with cognitive perfor-mance. The pattern of the EEG findingsindicated that power in the #{244}band (0.5 to3.5 Hz) was higher for subjects with loweriron status, possibly indicating less alertnessduring the experimental procedure. In ad-dition, many of the relationships of EEGpower with iron status were asymmetric,with less power (greater activation) in theleft hemisphere associated with higher serumferritin and iron levels. Although the smallsample size of this study causes any infer-ences to be tentative, a second cross-sec-tional study with evoked potential measuresalso indicated that left hemisphere activa-tion, reflected by higher amplitude evokedpotentials in left posterior electrodes, waspredicted by higher serum ferritin (12). In asubsequent 6-month longitudinal study ofseven men ( 1 3), we examined whether EEGand cognitive performance measures wouldreflect the decrement in iron status causedby repeated blood sampling. Although thefindings showed substantial individual vari-ability, and the t5 frequency data were not aspredominant as in the earlier cross-sectionalstudy, in four of the seven men asymmetricregression coefficients from the two hemi-spheres were observed that were consistentwith the previous cross-sectional study.

The major difficulty with these initialstudies is their small sample sizes. It appearsthat the patterning of the EEG data is im-portant to iron status, but to perform themultivariate analyses required to determinethis patterning with any confidence requiresa larger sample. The present research wasconducted to examine further the relationbetween iron status and cortical activationduring cognitive effort, to find if the resultsof the exploratory studies could be repro-duced and thus be generalized to the popu-lation. An important question was whether

measures of cognitive performance abilitywould show a relationship to iron status.Although it may be argued that electrophys-iological measures are more sensitive thanperformance measures, an absence of aneffect on psychological performance wouldcast doubt on the practical significance ofthe relation between cortical activation andiron nutriture observed in the earlier studies.In addition, because the earlier findings per-tam to iron status in normal persons andnot to iron deficiency, an important issue iswhether indices ofiron status may also indexsome other physiological attribute that coy-aries with brain function. Since animal stud-ies have suggested that hepatic ferritin isinfluenced by thyroid activity (14), and thy-roid function in turn influences corticalarousal ( 1 5, 16), we examined measures ofthyroid function in the present study to findif these might provide an alternative expla-nation for the apparent relationship betweeniron status and brain function.

Materials and methods

This study was approved by the human studies com-mittee of the University of North Dakota and theUnited States Department of Agriculture. Universitystudents were recruited from introductory psychologycourses for research on nutrition and brain function.The students were offered 10 dollars and course creditfor participation. Usable data were collected on 17males and 52 females. Since our previous studies mdi-cated a relation between iron nutriture and brain later-alization, e.g. Tucker & Sandstead (I I, 12), only right-handed subjects were recruited. When the subject re-ported to the laboratory a trained technician appliedthe EEG electrodes and data were collected during a30-5, eyes-closed resting period and while performingthe battery of eight cognitive tasks described below.During the same week as the EEG recording, a morningblood sample was taken to assess each subject’s char-acteristic level ofiron, serum ferritin, and several indicesof thyroid function. Subjects were required to fast afterthe previous day’s evening meal until the blood draw.

Cognitive tasks

Six cognitive tasks were performed with the eyesclosed. In a task emphasizing imagery. the subject wasasked to maintain a clear visual image of a concreteobject, such as a lawnmower, for 30 s and then rate thevividness of the image on a I to 7 scale. The wordfluency task required that the subject generate fourwords beginning and ending with specified letters foreach of six 20-s trials. The pair of letters was verballypresented and the subject pushed a button when readyto respond; performance measures were mean latencyto generate the four words and the total number of

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

IRON AND BRAIN FUNCTION 107

correct words. In none of the tasks was EEG datacollected during the production of a response, in orderto avoid movement or other artifact.

Additional eyes-closed tasks included digit span for-ward and backward and tonal memory forward andbackward. The digit span task followed the proceduresof the digit span subtest of the Weschler Adult Intelli-gence Scale: the subject was presented with a string of3 digits at a rate of 1/s. then after a brief pause wasrequested to repeat the digit string. EEG data werecollected in 1-s epochs as the subject listened to thedigit string and for the pause interval after the stringwas presented. After correct performance with a 3-digitstring, a 4-digit string was presented, with sequentiallylonger strings given until failure on two trials succes-sively. For the 3- and 4-digit trials 6 s of EEG data werecollected, with 8 5 collected for 5- and 6-digit trials, 105 for the 7- and 8-digit trials, and 12 s for the 9-digittrial if the subject performed sufficiently well to reachthat point. For the digit span backward task, the subjectwas requested to repeat the digit string in reverse order.The presentation was parallel to digits forward, and thenumber of seconds per trial was equal to twice thenumber of digits. Since the digits were presented at arate of 1/s. there was an equal number of seconds asdigits after the presentation before the subject was sig-naled to respond. Most people take longer to respondto digits backward than digits forward because of theadditional processing required to organize the response.

For the tonal memory forward task, a sequence of 2tones was presented, followed by a short pause, then asequence of 2 more tones; after a short interval thesubject was asked whether the 2 sequences oftones werethe same or different. After a second 2-tone trial, addi-tional tones were added sequentially with 2 trials ateach tone sequence length, up to 6 tones. For the 2-and 3-tone trials, 6 s of EEG data were collected; forlonger sequences the number of sec of EEG was twicethe number of tones. For the tonal memory backwardtask the subject was presented with 2 sequences of tonesand asked if the second sequence was the same ordifferent than the first sequence played backward. Thenumber of tones per sequence was increased incremen-tally as with the tonal memory forward, but the longestsequence presented for tonal memory backward was 5tones rather than 6.

Two additional cognitive tasks were performed withthe subject’s eyes open. In both the character and shapetasks, a group of letters was displayed in a pattern on avideo screen, and the subject was asked to attend eitherto the identity of the individual characters or the shapeof the pattern formed by those characters. After a 2-sinterstimulus interval, a second group of letters wasdisplayed and the subject responded whether the twogroups were the same or different, with respect to theattribute attended. Performance measures for the char-acter and shape tasks were the number of trials onwhich the subject was correct in his or her response,and the total latency of correct responses.

EEG recording and analysis

EEG signals were recorded from left and right frontal(F3, F4), temporal (T3, T4), parietal (P3, P4), andoccipital (01, 02) locations, referenced to linked ears

with a vertex ground. After amplification with a 0.1time constant and analog filtering with a 30 Hz b-passfilter the signals were digitized at 5 12 Hz and then 4-point smoothed to yield an effective sampling rate ofI 28 Hz. One-sec epochs of the raw data were thenexamined off-line on a display screen and examined forcontamination by eye movement, blink, myographic orother artifact. All EEG data collected during the cog-nitive tasks and the resting condition that were artifact-free were analyzed. After appropriate tapering of theepoch, spectral analysis with the fast Fourier transformproduced power spectra with 1.0 Hz resolution. Thisanalysis delineates the power (amplitude squared) inthe EEG waveform at each ofthe frequencies of interest.An example of power spectra computed for successiveepochs for a single subject is shown in Figure 1 . Foreach ofthe t5 (1 to 3 Hz), 0 (4 to 7 Hz), a (8 to 12 Hz),� 1 (l3toHz) bands, mean power for the band was computed foreach of the electrode sites for each cognitive task foreach subject. A complete description of the EEG anal-ysis techniques is provided elsewhere (1 1).

Assessment ofiron status and thyroid function

Serum femtin was assayed by competitive bindingradioimmunoassay. Iron was determined by atomicabsorption spectrophotometry (17). The samples wereaspirated directly after dilution with 2% HCI and werecompared with standards made up in HCI. Thyroidstimulating hormone was assayed by solid phase secondantibody techniques. Thyroxine (T4) and triiodothy-romine (T3) were measured by double antibody ra-dioimmunoassay.

Statisiica/ ana/.vsis

The covariation between individual measures of ironstatus and thyroid function, and between those meas-ures and cognitive performance, were examined withzero-order Pearson correlations. Canonical correlationswere used to determine the relationship between themultiple measures of iron nutriture and the multiplemeasures ofthyroid function. Canonical analysis simplyfinds a linear combination for each set of measures, thetwo resulting linear equations referred to as canonicalvariables, such that the zero-order correlation betweenthe two canonical variables is maximized. The distri-butions of the thyroid measures were positively skewedwith most values moderately low and a few values quitehigh. Log transformation of these distributions beforestatistical analysis yielded distributions that closely ap-proximated the normal.

Due to the positive skew characteristic of distribu-tions of EEG power measures, the EEG data were alsolog transformed before analysis. For each cognitive taskand each frequency band ofthe EEG, stepwise multipleregressions were run with the eight channels ofthe EEGas predictors and the measure of iron status (eitherserum ferritin or serum iron) as the criterion. A set ofregressions was also run with each of the thyroid meas-ures as the criterion, to determine ifthese would parallelthe regressions of EEG data on iron status. Finally, foreach cognitive task, regressions were run with the EEGdata from that task predicting task performance.Regression analysis forms a linear equation combining

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

F 4

108 TUCKER ET AL

FIG 1. An example of power spectra computed from successive epochs of 8 channels of EEG. The plot foreach epoch indicates the power at each of the frequency intervals resolved by the analysis. The increments on thex axis represent 5 Hz frequency intervals. The successive spectra are plotted with a hidden line routine to preventoverwriting previous plots. The peak in the spectra at about 10 Hz reflect the a frequency of the EEG.

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

IRON AND BRAIN FUNCTION 109

the various predictors in a least squares fashion suchthat the weight or coefficient assigned each predictorreflects its contribution to the prediction model. Thelinear equation is then used to predict the criterionvariables. The stepwise routine used adds predictors tothe equation, one at a time, generating the best one-predictor model. two-predictor model, etc. until allpredictors are included in the prediction ofthe criterion.In the analyses reported here, the most parsimoniousmodels were selected to characterize the EEG power-criterion (measures ofiron status, thyroid function, andcognitive performance) relationships. Thus, the modelschosen invoked the fewest predictors while still account-ing for most ofthe variance explained by the full eight-variable (eight channel) models.

Results

The mean serum ferritin level of the sub-jects in this sample was 4 1.8 ng/ml (SD =

28.4). The mean serum iron was 93.8 �tg/dl(SD = 33.0). As in previous research ( 1 1,12) the serum iron and serum ferritin meas-ures ofiron status were related (r = 0.51, p< 0.001).

EEG and iron status

The regressions of EEG data on each ofthe iron status measures showed strongerrelations to ferritin than to iron. Of the 48regressions run on ferritin (six EEG fre-quency bands for each of eight cognitivetasks), 39 were significant at the 0.05 levelor better (see Table 1). The variance ex-plained by the most parsimonious modelsranged from 0.07 to 0.29. As is shown bythe regression coefficients in Table 1, therewas a definite patterning of the EEG datathat predicted the subject’s serum ferritinlevel. Of the 39 left hemisphere predictorsretained in the significant models, 37 hadnegative signs, indicating less left hemi-sphere power (greater functional activation)was predictive ofhigher serum ferritin levels.For right hemisphere leads, greater power inthe temporal lobe was especially importantin the prediction of femtin level. Less powerin both the left and right occipital regionswas related to higher femtin. The right par-ietal location showed a frequency-specificityin the ferritin regressions, with less power inlower bands and more power in higher fre-quencies related to higher femtin levels.When these analyses were inspected for eachcognitive task, no clear differentiationamong the findings for each task was oh-

served, with results for the resting conditionsimilar to those for the cognitive perfor-mance conditions.

Because of the substantial sex differencesin ferritin level (with females coded as 0 andmales as 1 the correlation with serum femtinwas 0.46, p < 0.00 1), separate regressionswere run for the male and female samples.Although somewhat more significant find-ings appeared for the males, both sexes re-flected the patterning of the results of thecombined analysis, thus indicating that gen-der differences could not account for therelation between EEG data and ferritin level.

The regressions of EEG power on serumiron showed only 1 1 ofthe 48 regressions toyield significant (p < 0.05) models (see Table2). Of these, however, some specificity wasobserved; five of the significant models werefor the #{244}frequency with the squared multiplecorrelations ranging from 0. 10 to 0.25. Aswith the ferritin regressions, we examinedthe zero-order correlations between the EEGpredictor channels and the iron criterionmeasure to find if the signs of the regressioncoefficients could be interpreted to reflectdirect relations with the criterion (ie, thecoefficients were not due to suppressor ef-fects in the regression analysis). Particularlyfor the frontal lobes, the asymmetry of theregression coefficients was consistent acrossseveral tasks, with less power in the leftfrontal and greater power in the right frontallocations related to higher serum iron. Incontrast to the results of our first cross-sectional study (10), there was no particulartendency for serum iron to covary with #{244}power recorded during the resting condition.

Iron status and cognitive performance

The relations between the iron measuresand performance of the cognitive tasks wereexamined after controlling for gender differ-ences in cognitive task performance (whichwere significant for the digit span backward,tonal memory forward, shape, and charactertasks). These partial correlations showedhigher serum ferritin to be associated withpoorer performance of the tonal memorybackward task (r = -0.30, p < 0.05). Serumiron was associated with better (shorter Ia-tency) performance on the word fluency task(r = -0.30, p < 0.05).

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

TABLE 1Multiple regressions of EEG power on serum ferritin separatelyfor each frequency band and recording conditionsignificant (p < .05) models only

‘5

Fee-quencytask

F3

Regression coefficients

Channel

F4 T3 T4 P3 P4 01 02 R2

-78

-52

-19

-88 111

46

-30

36

-40

1/�

DSFDSBTMFWF52C2

a

DSFDSBWF52C2

R1MGDSFDSBTMFTMBWF52C2

�i2R1MGDSFTMFTMBWFC2

-48

-41

-39

-66

-48

-35-34

-23-24

-5142 -7754 -7154 -8141 -6146 -65

-76-49

65 -55

36 -59

-39

-47 803872

-36 54

56

-21 42

274231

1819

21

-22 302317

-27 28-29 23

15

II-12

0.20t0.14*0.16*0.09*0.07*

-43 0.l7t0.28t0.29�0.l6t0.2lt0.18*0.19t0.22t0.l8t

-27 0.15*0.26t0.15*0.17*0.12*0. 11�

-27 0.08*

-24-50-26-32-34-30-35

I 10

R1MGDSFDSBTMFTMBWF52C2

1:13R1MGDSFDSBTMFTMBWF*p<0.05.

t p < 0.01.:1:p < 0.0001.§ p < 0.001.

28

24

15-48 37

25

-47 65

34

78 -53-75 152-59 122

4423

-37 0.10*0.1 1*

-60 0.l9t0. 19t0.20t

-48 0.l6t-49 0.12*-31 0.24t-29 0.16*

0.24t-44 0.13*

0.12*0.13*0.26t0.10*

-21 0.14*0.29�

-44 0.22*-29 0.17*

0.12*0.2lt

_______0.12*

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

TABLE 2Multiple regressions of EEG power on serum iron separatelyfor each frequency bland and recording conditionsignificant (p < 0.05) models only

Fre-quency

task

Regression coefficients

Channel

F3 F4 T3 T4 P3 P4 01 02 R2

‘5DSF -36 55 -64 0.17*DSB -65 46 77 -102 0.25tTMF 41 -70 0.11*TMB -71 61 72 -62 0.19*S2 -127 71 0.22t

0

S2 56 -174 96 0.17*

a

R 64 -49 0.12*1MG 59 -26 0.12*WF 53 -66 23 0.16*

01

�321MG 38 0.10*

03TMF -48 33 0.14**p.<o.05.

t p < 0.01.

IRON AND BRAIN FUNCTION Ill

These findings suggest that cognitive abil-ity is related to iron nutriture in a way thatis consistent with the asymmetries found inthe EEG predictions ofthe iron indices: boththe activation and the cognitive perfor-mance of the left hemisphere are related toiron status. Although the EEG findings arenot as consistent for the right hemisphere,they are still generally congruent with thecognitive data: lower right hemisphere elec-trophysiologic activation (with the exceptionof the occipital lobe) and poorer nonverbalcognitive performance were associated withhigher levels of iron stores. The third leg ofthe triangle relating iron status, brain elec-trophysiology, and cognition describes therelations between the EEG data and cogni-tive performance. Although too extensive toreport here, these analyses indicated that lefthemisphere activation (decreased power)was significantly predictive of shorter re-sponse latency for the word fluency task.Thus left hemisphere electroph�siologicalactivation is relevant to verbal cognition,

and may mediate the relation between ironstatus and cognitive performance.

Iron status and thyroid function

The mean thyroid stimulating hormonelevel for this sample was 4.07 �dU/ml (SD= 1 .7). The mean for triiodothyromine was135.4 ng/mI (SD = 4 1.2) and the mean forthyroxine was 8.0 ng/ml (SD = 2.9). Underthe hypothesis that thyroid activity mediatesthe apparent relation between EEG meas-ures and iron status it would be expectedthat serum ferritin or iron levels would beclosely related to thyroid activity. The can-onical correlation between iron status(serum iron, ferritin) and thyroid (thyroidstimulating hormone, triiodothyromine,thyroxine) variables showed no significantassociation between iron status and thyroidfunction in this sample. Similarly, zero-or-der correlations between the individual ironstatus measures and the thyroid variableswere low and nonsignificant.

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

112 TUCKER ET AL

Discussion

The results of this research indicate thatindividual differences in iron status are rel-evant to specific aspects of brain function.The asymmetries in the EEG data in thissample generally replicate those in our ear-her studies (10- 12), and are probably char-acteristic of the population at large. Al-though the measures of cognitive perfor-mance in this study were fairly crude, ironstatus was related to performance in a man-ner consistent with the EEG results. In theseright handers greater left hemisphere electro-physiological activation and better wordfluency performance were associated withhigher iron status, while less activation ofthe anterior right hemisphere and poorerperformance on a nonverbal task were char-acteristic of higher iron status.

Since few ofthe subjects in this study wereiron deficient, it is not certain that the find-ings reflect a relation between iron statusand brain function that is easily influencedby dietary intake ofiron. Although the anal-yses of the thyroid measures indicate thatthyroid function cannot serve as an alter-native explanation for the relation betweenEEG and iron status, other factors may beimportant as mediating variables. For ex-ample, evidence is now impressive that lowlevels of lead measurably impair cognitivefunction and alter brain electrophysiology( 18, 19). Animal studies show that dietaryiron deficiency can result in increased ab-sorption of several metals, including lead, aswell as iron (20). Low iron status might thusincrease the risk of lead exposure. In thefuture the possibility of lead and other met-als confounding the results of studies of re-lations between iron nutriture and neuro-psychological function needs to be evalu-ated.

Perhaps the most interesting aspect of therelation between iron status and the neuro-psychological measures in this study is thespecificity of the relationship. What ac-counts for the particular importance of ironto the left hemisphere? Given the evidencethat adequate iron nutriture is required fornormal functioning of the dopamine neu-rotransmitter pathways in the brain (6), itmay be relevant that recent evidence sug-

gests that dopaminergic pathways are partic-ularly important to the activation and func-tion of the left hemisphere in humans (21).In a series of studies of iron deficiency inrats, Youdim and associates (6, 7, 22) haveshown that dopamine receptor functionseems to be the mediating factor betweeniron deficiency and behavioral impairment.Iron is highly concentrated in brain dopa-mine pathways (8). Iron-deficient rats showdecreased behavioral effects ofdopamine ag-onists, such as amphetamine, and changesin brain iron occur after administration ofdopamine antagonists (22). Dopamine an-tagonists have been found to influence lefthemisphere EEG activity specifically in bothschizophrenics (23) and normal subjects(24). In psychiatric patients, levels of a do-pamine metabolite (homovanillic acid) inthe cerebral spinal fluid were found to cor-relate with cortical evoked potentials fromthe left but not right hemisphere (25). Inlight of the suggestion by Ashkenazi et al(22) that a decrease in brain iron turn-overcaused by neuroleptics (dopamine antago-nists) may result in the tardive dyskinesiasproduced by these drugs, it is interesting thattardive dyskinesias are more frequent on theright side ofthe body (26). Ifiron is requiredto support adequate dopamine neurotrans-mission, the direction ofthe present findingswould be consistent with the apparent left-lateralization of dopaminergic pathways inhumans.

In the regressions of EEG power on theindices of iron status, it was not only lefthemisphere signals but the asymmetry of theEEG, including relatively less right hemi-sphere activation, that was most predictiveof iron status. Furthermore, the perfor-mance on the tonal memory backward taskby subjects with higher ferritin levels suggestpoorer right hemisphere function as well asdecreased electrophysiological activation.Given reciprocal inhibition between thehemispheres, a change in iron nutrituremight alter asymmetric neurotransmittersystems and thus influence the relative bal-ance between the two hemispheres’ func-tioning. Alternatively, whatever genetic anddevelopmental factors predispose an mdi-vidual to rely on one hemisphere more thanthe other for cognition might also cause a

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from

IRON AND BRAIN FUNCTION 113

greater need for the biochemical substratesupporting that hemisphere’s operations. Inthat case, iron nutriture would be the de-pendent variable in the equation, with theindividual’s biochemical characteristics dic-tating the iron requirement.

Although there are a number ofissues thatmust be better understood to explain theconnection between iron nutriture and brainfunction, the present study provides furtherevidence that the level of body iron, and byimplication brain iron, is associated withqualitatively specific aspects ofbrain electro-physiology that are in turn important tocognitive function. Clarification of the im-plications of these relationships for personswith iron deficiency and for persons judgediron adequate by usual standards awaits fur-ther research. a

The authors thank Bruce Sawler for technical sup-port, Sharon Shade and Elizabeth Lucas for data collec-tion, Sandy Gallagher for clinical chemistries, Tom Bairfor systems programming, and the students ofthe Uni-versity of North Dakota who participated in this re-search.

References

1. Fowler WM. Chlorosis-an obituary. Ann MedHistory l936;8: 168-77.

2. Control ofnutritional anemia with special referenceto iron deficiency. Technical report series 580. Ge-neva: WHO, 1975.

3. Sulzer JL, Wesley HH, Honig F. Nutrition, devel-opment and social behavior. Washington, DC:DHEW publication 73-242, 1973.

4. Webb TE, Oski FA. Iron deficiency anemia andscholastic achievement in young adults. J Pediatr1973;82:827-30.

5. Pollitt E, Greenfield D, Leibel R. Significance ofBayley scale score changes following iron therapy.J Pediatr 1978:92:177-8.

6. Youdim MBH, Green AR. Iron deficiency andneurotransmitter synthesis and function. Proc NutrSoc 1978:37:173-9.

7. Youdim MBH, Yehuda S. Ben-Shachar D, Ash-kenazi R. Behavioral and brain biochemicalchanges in iron deficient rats: involvement of ironin dopamine receptor function. In: Pollitt E, LeibelR, eds. Iron deficiency, brain biochemistry andbehavior. New York, NY: Raven Press, 1982.

8. Sourkes TL. Transition elements in the nervoussystem. In: Pollitt E. Leibel R. eds. Iron deficiency,

brain biochemistry and behavior. New York, NY:Raven Press, 1982.

9. Kahneman D. Attention and effort. EnglewoodClifts, NJ: Prentice-Hall, 1973.

10. Pribram KH, McGuinness D. Arousal, activationand effort in the control of attention. Psychol Rev1975:82:1 16-49.

1 1. Tucker DM. Sandstead HH. Electroencephalo-graphic correlates of iron status: Tired blood revis-ited. Physiol Behav 1981: 26:439-49.

12. Tucker DM, Sandstead HH. Body iron stores andcortical arousal. In Pollitt E, Leibel R, eds. Irondeficiency, brain biochemistry and behavior. NewYork, NY: Raven Press, 1982:162-81.

13. Tucker DM, Sandstead HH, Swenson RA, SawlerBG, Penland JG. Longitudinal study ofbrain func-tion and depletion of iron stores in individual sub-jects. Physiol Behav l982;29:737-40.

14. Winkelman JC, Mariash CN, Towle HC, Oppen-heimer JH. Thyroidectomy increases rat hepaticfemtin iron. Science 1981:213:569-71.

15. Leonard JL, Kaplan MM, Visser TJ, Silva JE,Larsen PR. Cerebral cortex responds rapidly tothyroid hormones. Science 1981:14:571-3.

16. Marwaha J, Prasad KN. Hyperthyroidism elicitselectrophysiological noradrenergic subsensitivity inrat cerebellum. Science 1981:214:675-77.

17. Olson AD, Hamline WB. A new method for serumiron and total iron binding capacity. Clin Chem1966; 12:47.

18. Needleman HL, Gunnoe C, Leviton A, et al. Def-icits in psychologic and classroom performance ofchildren with elevated dentine lead levels. N EngIJ Med 1979:300:689-95.

19. Thatcher RW, Lester ML, McAlaster R. Horst R.Effects of low levels of cadmium and lead on cog-nitive functioning in children. Arch Environ Health1982:37:159-65.

20. Flannigan PR, Haist J, Valberg LS. Comparativeeffects of iron deficiency induced by bleeding anda low-iron diet on the intestinal absorptive inter-actions of iron, cobalt, manganese, zinc, lead andcadmium. J Nutr 1980:1 10:1754-63.

2 1. Tucker DM. Lateral brain function, emotion andconceptualization. Psychol Bull 198 1;89: 19-46.

22. Ashkenazi R, Ben-Shachar D, Youdirn MBH. Nu-tritional iron and dopamine binding sites in the ratbrain. Pharmacol Biochem Beh 1982:l7(suppl 1):43-7.

23. Serafetinides EA. Voltage laterality in the EEG ofpsychiatric patients. Dis Nerv Syst 1973:34:190-1.

24. Laurian S. Gaillard JM, Le PK, Shopf J. Topo-graphic aspects of EEG profile of some psycho-tropic drugs. Adv Biol Psychiat (in press).

25. Gottfnes CG, Pert-is C. Roos BE. Visual averagedevoked responses (AER) and monoamine metabo-lites in cerebral spinal fluid (CSF). Acta PsychiatrScand 1974:(suppl 255):l35-42.

26. Waziri R. Lateralization of neuroleptic-induceddyskinesia indicates pharmacological asymmetry inthe brain. Psychopharmacologia 1980:68:51-3.

by guest on March 20, 2014

ajcn.nutrition.orgD

ownloaded from