Am J C/in Nuir 1990:52:8 13-9. Printed in USA. © 1990 American Society for Clinical Nutrition 813

Impaired thermoregulation and thyroid functionin iron-deficiency anemia�3

John L Beard, Myfanwy J Bore!, and Janice Derr

ABSTRACT Ten women with iron-deficiency anemia, 8

with depleted iron stores (nonanemic), and 12 control women,

all of similar body fatness, were exposed to a 28 #{176}Cwater bathto test the hypothesis that iron-deficiency anemia impairs ther-

moregulatory performance. The anemic women had lower rec-

tal temperatures than did control women (36.0 ± 0.2 vs 36.2

± 0. 1 #{176}C,respectively, P = 0.001) and a lower rate of oxygenconsumption (5.28 ± 0.26 vs 5.99 ± 0.29 mL.min’ .kg body

wt�’, respectively, P = 0.04) at 100 mm of cold exposure.

Plasma thyroxine and triiodothyronine concentrations were

significantly (P < 0.002) lower in anemic than in controlwomen at baseline and during cold exposure. Responses of

iron-depleted subjects were similar to those ofcontrol subjects.Iron supplementation corrected the anemia, significantly (P

= 0.03) improved rectal temperature at 100 mm, and partiallynormalized plasma thyroid hormone concentrations. Plasmacatecholamines were unaffected by iron status. This ex-periment demonstrates a functional consequence of iron-defi-ciency anemia in the balance of heat production and loss andsuggests that thyroid-hormone metabolism may be respon-

sible. AmfClinNutr 1990;52:813-9.

KEY WORDS Iron deficiency, temperature regulation,

human subjects, plasma catecholamines, thyroxine, triiodo-

thyronine

Introduction

Iron deficiency continues to be a significant nutritional-de-

ficiency disease in many parts of the world. According to esti-mates by the World Health Organization, ‘� 15% ofthe world’s

population has significant iron-deficiency anemia (1). In theUnited States, prevalence estimates from the second NationalHealth and Nutrition Examination Survey (NHANES II) mdi-cated that 1-6% ofthe population has impaired iron status (2).Numerous metabolic consequences of iron deficiency havebeen described (3-7). In the rat model, poor thermoregulation

is one of these deleterious consequences (8-1 1), with indica-

tions that thyroid-hormone metabolism and catecholaminemetabolism are both significantly altered in iron deficiency

(8-13).

One preliminary report in humans (14) showed that iron-deficient-anemic subjects failed to thermoregulate normally

during cold-water immersion, with a resultant significant loss

of body temperature. Interpretation of that study was con-founded by significantly lower body fat, as indicated by height-to-weight ratio, in iron-deficient-anemic compared with con-

trol subjects. Thus, the significantly greater metabolic rates andhigher plasma norepinephrine concentrations in the anemic

subjects could have been causally related to both their body

fatness and their iron deficiency.As an extension ofour interest in alterations in thermoregu-

lation during cold exposure in iron deficiency, we wereprompted to use cold stress as a useful probe to examine how

control of energy metabolism is perturbed in iron deficiencyand, more specifically, how the hormones that control heat pro-duction and metabolic rates are affected by this single nutrientdeficiency. We thus chose to examine thermoregulatory perfor-

mance in young women with iron-deficiency anemia during amoderate cool-water immersion and to determine if alterationsin thyroid hormone and catecholamine metabolism might be

the underlying metabolic defects once the body fatness of sub-jects was strictly controlled. Moreover, we proposed to see if

these alterations were repaired and reversed with oral iron sup-plementation.

Methods

Research design

This research was designed as a combined cross-sectional

and longitudinal study using a pre- and posttreatment experi-

mental design with oral iron supplementation as the treatment.Such a design allowed us to examine differences in thermoregu-latory capacity between iron-deficient anemic (IDA), iron-de-

pleted (IDP), and iron-sufficient control (C) women of similarpercent body fat (cross-sectional comparison) and in IDA andIDP women before and after oral iron supplementation (longi-

tudinal comparison). During the presupplementation phase ofthe study, thermogenic performance was assessed while eachsubject underwent two or three cool-water-bath exposures.

Multiple cool-water exposures were performed on each subjectbecause ofintraindividual variation in certain physiological re-sponses. After the initial exposures, the IDA and IDP women

I From the Departments of Nutrition and Statistics and the NollLaboratory for Human Performance Research, Pennsylvania StateUniversity, University Park.

2 Supported by a grant from the US Department ofAgriculture’s Hu-man Nutrients Requirement Program.

3 Address reprint requests to JL Beard, Nutrition Department, Penn-sylvania State University, S125-F Henderson, University Park, PA16802.

Received August 29, 1989.Accepted for publication November 21, 1989.

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

‘I

814 BEARD ET AL

�O2

CORE TEMP.BLOODDRAW

received 12 wk of oral iron supplementation (78 mg Fe/d as

ferrous sulfate), and the control women received no supple-

mentation. After the supplementation period, two to three

cool-water-bath exposures were repeated in all three groups ofsubjects to determine the impact of oral iron supplementation

on thermogenic performance in the IDA and IDP women.

Subject selection

Women were recruited for this study from The Pennsylvania

State University campus through the use ofannouncements inacademic classes, ads in the school newspaper, and flyers

posted around the campus. Potential subjects reported for aninitial screening, at which time the study protocol was fully ex-

plained, informed consent forms were signed, and a finger-prick blood sample was taken to assess the individual’s ironstatus from hemoglobin (Hb) and serum ferritin (SF) concen-trations. Women were selected for participation in the studyaccording to four criteria: 1) aged between 18 and 44 y; 2) freeofany known metabolic disorders such as diabetes, hypoglyce-

mia, or thyroid gland disorders as reported by the subject; 3)

percent body fat between 20% and 30% as determined by hy-drostatic weighing, including a nitrogen wash-out technique toestimate residual lung volume (15); and 4) iron status, withsubjects classified as iron-deficient-anemic (Hb < 120 g/L and

SF < 12 �g/L), iron-depleted (Hb > 120 g/L and SF < 12 �g/L), or iron-sufficient control (Hb > 120 g/L and SF > 20 �g/

L). On the basis of these criteria, 10 IDA, 8 IDP, and 12 Cwomen having a similar normal percent body fat were selected

for the study. All subjects underwent a complete medical exam-ination (medical history, electrocardiogram, blood pressure,

and physical exam) by a physician before their participation inthe study. The protocol for this study was approved by theOffice for the Protection of Human Subjects at The Pennsylva-

nia State University.

Water-bath data collection

On the day of the cool-water-bath exposure, subjects re-ported to the Noll Laboratory for Human Performance Re-search on the Penn State campus. The subject’s weight, time oflast meal, and day of the menstrual cycle were recorded. Sub-

jects fasted � 5 h before the water bath to avoid the contribu-

tion ofdietary thermogenesis to metabolic-rate measurements.

In addition, the water-bath treatments were performed duringthe follicular phase (first 14 d) ofthe menstrual cycle to avoidthe confounding effect ofthe increase in body temperature andmetabolic rate that occurs during the luteal phase (last 14 d) ofthe cycle (16).

A nurse placed a 20-gauge catheter(Abbocath, Abbott Labo-ratories, Chicago) in the antecubital vein of the subject’s right

arm to allow for periodic blood sampling during the water bath.After placement of the catheter, the subject entered a large

polypropylene tank filled with warm water (35 ± 1 #{176}C)and sat

in a chair so that the water level was covering her shoulders.

The subject’s right arm was placed on an armrest out of thewater to prevent water from contacting the site of the catheterplacement. The rectal thermister was connected to a telether-

mometer (Yellow Springs Instruments, Yellow Springs, OH)to measure core body temperature (±0. 1 #{176}C).A face mask (Re-

spironics, Inc. Monroeville, PA) was placed over the subject’snose and mouth to allow for the continuous collection of ex-pired respiratory gases for oxygen consumption and metabolic-

,� ,� 9 � ,�

I I I � I I

BASELINE � 0’ 20’ 40’ 60’ 80’ 100’

35#{176}C�2WR 28#{176}C

FIG 1. Timeline representation of the water-bath data-collection

procedure.

rate measurements. After the face mask was in place, the lightswere turned down and the subject was instructed to relax in the

warm water. This warm-water baseline period lasted for either30 or 45 mm, depending on how long it took the subject’s re-spiratory-gas measurements to stabilize at a baseline level. At

the end of the baseline period, the nurse withdrew a lO-mLblood sample for later hematological and hormonal analyses,and core body temperature and respiratory-gas measurementswere recorded.

At the end ofthe warm-water baseline period, 5 #{176}Ccold waterand ice were added to the tank to mix with and replace thewarm water and to quickly (10-1 5 mm) lower the temperature

ofthe water bath to 28 #{176}C.A pump system circulated the coolwater throughout the tank to ensure a uniform water tempera-ture. The subject remained in this cool water for 100 mm. At

the time the water reached 28 #{176}C(time 0) and at 20-mm inter-vals thereafter until 100 mm, a blood sample was drawn andcore body temperature and respiratory-gas measurements were

recorded. Figure 1 presents a timeline overview of the water-bath data-collection procedure. To facilitate the subject’s abil-ity to remain in the cool water for 100 mm, she was shown amovie on a videocassette recorder.

Blood analyses

Each subject’s iron status was assessed from Hb and SF con-

centrations from a capillary blood sample (17) during thescreening procedure and from a venous blood sample duringthe cool-water-bath exposure. Hb and SF concentrations weremeasured colorimetrically by use of a cyanmethemoglobinmethod (Sigma Chemical Company, St Louis) and an enzyme-linked immunosorbent assay (ELISA) technique (18), respec-

tively.

The lO-mL venous blood samples drawn at 20-mm intervalsduring the cool water bath were equally partitioned into hepa-rinized tubes for later hematologic and hormone analyses andinto tubes containing EGTA and glutathione (AmershamCorp, Chicago) for catecholamine determination. The blood-

collection tubes were kept on ice until they were centrifuged at1 350 X g for 12 mm to separate the plasma from the red blood

cells. The plasma was then frozen for later analysis of triiodo-thyronine (T3), thyroxine (T4), estrogen, progesterone, thyroid-

stimulating hormone (TSH), epinephrine, and norepineph-rime. Plasma samples were stored at -20 #{176}Cuntil later analysisexcept for the catecholamine samples, which were stored at-80#{176}C.

Plasma T3 , T4 , and TSH as well as estrogen and progesteronewere measured by radioimmunoassay (kits from Monobind,

Costa Mesa, CA, and Serono Diagnostics, Inc, Norwell, MA,respectively). The coefficient of variation in all assays was

< 6%. Plasma catecholamine samples were extracted and epi-

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

THERMOREGULATION AND IRON DEFICIENCY 815

and after iron supplementation.

nephrine and norepinephrine concentrations were determined

by use of high-pressure liquid chromatography with electro-

chemical detection (HPLC-EC) according to described meth-

ods (19). A refrigerated autosampler (Waters 7 12 WISP, Milli-pore Corp, Milford, MA) injected the samples, and an integra-

tor (RC3-a, Shimadzu Corp, Tokyo) was used to determinepeak areas under the curve. Plasma epinephnne and norepi-

nephrine concentrations were calculated based on a five-point

standard curve and were corrected for recovery of the internal

standard, dihydroxybenzylamine. The average recovery of the

internal standard was 67%.

Respiratory-gas measurements

Continuous respiratory-gas measurements were made dur-ing the cool water bath by use of open-circuit methods. As de-

scribed earlier, subjects wore a face mask (Respironics, Inc),which covered their nose and mouth. Sufficient continuous airflow was maintained through the mask to prevent the accumu-

lation of carbon dioxide in the mask, and the flow rate wasrecorded. Oxygen and carbon dioxide concentrations of ex-pired air were measured with an oxygen analyzer (Applied

Electrochemistry S3-A, Sunnyvale, CA) and an infrared carbon

dioxide analyzer(LB-2, Beckman, Fullerton, CA), respectively.

A dual-channel chart recorder provided a continuous record of

the oxygen and carbon dioxide concentrations of the subject’s

expired air. The respiratory quotient (RQ), expired air volume!mm (VE), and rates ofoxygen consumption (V02) and carbondioxide production (VCO2) were calculated for the baseline pe-riod and the 20-mm time points during the water bath by use

ofstandard equations(20). Planimetry was used to measure thearea under the oxygen curve, and the total volume (L) of oxy-

gen consumed by the subject during the 100 mm of cool-water-bath exposure was calculated.

For the determination ofenergy expenditure during the cool-

water-bath exposure, we measured urinary nitrogen excretionin a post-water-bath urine sample in a subset of the subjects

and calculated energy expenditure according to Weir’s equa-tion (21). In addition, since the contribution ofprotein metab-

olism to overall energy metabolism is generally small (20), wecalculated energy expenditure assuming that the RQ was repre-

sentative of a nonprotein RQ. We found that energy expendi-ture during the cool water exposure as estimated by assuming

a nonprotein RQ without correction for the contribution ofprotein metabolism varied from that estimated from a true

nonprotein RQ (correction for urinary nitrogen excretion) byonly 0-2% (data not shown). Thus, we elected to calculate each

subject’s energy expenditure in kilocalories during the water

bath by multiplying the thermal equivalent ofoxygen (kilocalo-ries expended per liter ofoxygen consumed) (22), based on thesubject’s nonprotein RQ value (average ofRQ values at the 20-mm time points), by the total liters ofoxygen consumed during

the bath, as determined from planimetry.

Statisticalanalvses

The general analytic scheme was a two-way repeated-mea-sures analysis of variance (ANOVA) with iron group as the be-tween-subject factor and supplementation status as the within-

subject factor. There were two to three replicate water baths foreach subject at both pre- and postsupplementation. We decided

a priori to perform separate, cross-sectional analyses at the fol-lowing three time points: baseline and 40 and 100 mm of cool-

water exposure. The significance ofdifferences between meanswas determined by contrasts of the least-square means (LS

means) for a priori comparisons ofiron group by supplementa-

tion status. The level of significance used for each a priori testwasP= 0.05.

We performed separate ANOVAs for each of the response

measures. Before the analysis, we inspected the distribution of

each response measure for possible departures from the stan-dard ANOVA assumptions ofnormally distributed errors withconstant variance. This inspection led us to make the followingtransformations: 1) the logarithm of plasma epinephrine wasused as the dependent response in order to stabilize the error

variance and 2) the median ofthe two or three replicates at each

time sample was used as the dependent variable for plasma T4,T3 , and TSH and for VO2 (mL . min I � kg body wt ‘) in orderto decrease the influence of outlying data values. This latter

analysis was then weighted by the number ofreplicates contrib-

uting to each median. The distribution of the remaining re-sponse variables conformed satisfactorily to ANOVA assump-tions and did not require transformation.

Body composition data were analyzed by one-way ANOVA.All data presented in the figures are 1± SEM. The hematologicand body composition data are presented asi± SD. All statisti-

cal analyses were performed by use of the Statistical Analysis

System (SAS Institute Inc, Cary, NC).

Results

Our selection criteria based on SF and Hb concentrations

and body fatness resulted in three groups ofsubjects(IDA, IDP,and C) that differed initially only in iron nutriture (Table 1).Both iron-deficient groups had depleted iron stores, as mdi-

cated by low SF concentrations, and the control group was wellabove the cutoff value of 12 �g/L. Iron supplementation waseffective in correcting the anemia of the IDA group but not in

restoring the SF to normal concentrations. Supplementationalso significantly (P = 0.0001) improved the concentration ofSF in the IDP women. We had to reassign one subject with an

initial Hb concentration > 120 g/L from the IDP to the IDA

group because ofan increase of24 g/L in her Hb concentration

after iron supplementation. As expected from our screeningcriteria, there were no differences in percentage body fat, leanbody weight, or body surface area among groups (Table 2).

There were no significant differences among the groups withrespect to the number of hours fasted before the water bath(means for each group ranged from 8.8 to 10.8 h) or day of the

menstrual cycle (means ranging from day 5 to day 1 1 ). Plasmaestrogen and progesterone measurements (data not shown) ver-ified that water baths were performed while subjects were in the

follicular phase ofthe menstrual cycle.Exposure to 28 #{176}Cwater for 100 mm (Fig 2) resulted in a

significantly lower mean core body temperature in the IDAsubjects (36.0 ± 0.2 #{176}C)than in the C (36.2 ± 0. 1 #{176}C)or IDP

subjects (36.4 ± 0. 1 #{176}C).Iron supplementation resulted in a

significant improvement in the ability of the formerly anemic

subjects to maintain body temperature after 100 mm of cool-water exposure such that the IDA and C groups did not differ

significantly at 100 mm. Interestingly, the IDP subjects had a

significantly higher mean core temperature than did the IDA

or C subjects after 100 mm ofcool-water exposure, both before

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

-20 #{212} �0 �o �o lo i

816 BEARD ET AL

51±SD.

TABLE I

Hematologic characteristics of the subjects5

Control(n= 12)

Anemic(n= 10)

Iron de

(n=pleted

8)

Pre Post Pie Post Pre Post

Serum ferritin (pg/L) 34. 3 ± I 1.8#{176} 30.4 ± 14#{149}2b 3� 8 ± 2.4c 9.7 ± 5#{149}7d 5� 1 ± 35C 16 .3 ± 95C

Hemoglobin(g/L) 136 ± 9� 137 ± 7#{176} 110 ± 11 b 128 ± 13C 132 ± 6a.d 132 ±6d

S j: � SD. Pre. before supplementation with iron: post, after supplementation.Means with different superscripts within a row are significantly different (P < 0.05) for pre- vs post-comparisons within a hematologic-characteris-

tic group and for pre-comparisons and post-comparisons across groups.

VO2 ofthe three groups ofsubjects did not differ at baseline

before iron supplementation (Fig 3). The IDA subjects didhave a significantly lower VO2 than did the C subjects at 100mm and than the IDP subjects at both 40 and 100 mm of cool-

water exposure. After the supplementation period, the V02 of

the IDA women was significantly lower than that of the Cwomen at baseline and at 40 and 100 mm of cool-water expo-sure. Iron supplementation in the IDP women was associatedwith significant decreases in V02 at all three time points in the

postsupplementation water baths. Before supplementation, the

percentage rise in VO2 above baseline was significantly less in

the IDA than in the C women (64 ± 8% vs 86 ± 7%, respec-

tively; P = 0.04) at 100 mm as well as 40 mm (30 ± 8% vs

48 ± 7%, respectively; P = 0.03). These differences were not

significant at 40 mm but remained significantly different at 100mm ofexposure (68 ± 9% vs 82 ± 10%, respectively; P = 0.04)in postsupplementation comparisons. The mean RQ ranged

from 0.78 to 0.85 for the three groups and did not differ amonggroups except for a significantly lower RQ in the IDA groupat baseline, both before (P = 0.01) and after (P = 0.02) iron

supplementation when compared with the C women, and post-supplementation (P = 0.003) when compared to the IDPwomen.

We calculated the total VO2 of the subjects during the 100mm ofcool-water exposure and converted these data to kilocal-

ories expended during the water bath (Table 3). The IDAwomen expended ‘�8% fewer calories than did C women and

nearly 17% fewer than did IDP subjects before iron supplemen-

tation; the latter difference was statistically significant (P

= 0.0004). In addition, the mean kilocalorie expenditure of the

IDP women was significantly (P = 0.02) greater than that of theC women, presupplementation. Iron supplementation normal-ized the caloric expenditure ofboth iron-deficient groups such

that there were no significant differences among the three

groups.

In our examination of the potential underlying hormonal

TABLE 2Body composition characteristics ofthe subjects

Control(n= 12)

Anemic(n= 10)

Iron depleted

(n=8)

Percent body fat (%) 25. 1 ± 3.2 24.2 ± 3.0 25.6 ± 2.4

Lean body weight (kg) 45.0 ± 5. 1 46.7 ± 4.3 44.2 ± 3.2

Surface area(m2) 1.68 ± 0. 14 1.69 ± 0. 1 1 1.66 ± 0.09

changes associated with this poor thermoregulatory capacity

in iron deficiency, we focused on the thyroid hormones and

catecholamines. Figure 4 shows quite clearly that IDA subjectshad significantly lower plasma T3 concentrations than did the

C or IDP women throughout the water baths. These differencespersisted even after iron supplementation despite a significant12% rise in T3 concentration in the IDA group at baseline afteriron therapy. Women with depleted iron stores had signifi-cantly lower mean plasma T3 concentrations than did the Cwomen at baseline and at 40 and 100 mm of cool water expo-

sure after the supplementation period.As shown in Figure 5, plasma T4 concentrations were also

significantly lower in the IDA subjects than the C or IDP sub-jects throughout the water baths and were not normalized with3 mo of iron supplementation. Before iron supplementation,

the IDP women had a significantly greater mean TSH concen-tration than did the C women at baseline (data not shown).

This difference was not evident at the 40- or 100-mm time

points or after the iron-supplementation period.Plasma norepinephrine concentrations did not differ among

37.0

00

‘::;‘ 36.6

� 36.4

� 36.2

0�

uJI-

36.0

TIME (minutes)

35.8

FIG 2. Core temperature response of the 12 control (C), 10 iron-deficient-anemic (IDA), and 8 iron-depleted (IDP) women exposed to

a 28 #{176}Cwater bath before (pre) and after (post) iron supplementation.0 - 0, control-pre; #{149}- #{149}1 control-post; E� - L�., iron-dc-ficient anemic-pre; � - A, iron-deficient anemic-post; 0-0,iron-depleted-pre; U - �, iron-depleted-post. I ± SEM for base-line (- 10), 40, and 100 mm. Core temperature was significantly lower

in IDA than C(P = 0.00 l)and IDP(P = 0.000 1) at 100 mm presupple-mentation. Core temperature was significantly (P = 0.03) improved inIDA at 100 mm postsupplementation. Core temperature of IDP was

significantly higher than C(P = 0.02, pre; P = 0.000 1 , post)and IDA(P= 0.0001, pre and post) both pre- and postsupplementation at 100 mm.

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

TIME (minutes)

temperature during the current water immersion in our C andIDP subjects was less than that observed by McArdle et al (23)in four women ofsimilar body fatness. The lesser drop in body

temperature in our women is attributed to their lower baselinetemperature compared with the subjects ofMcArdle et al (23).

The relatively poor metabolic response of the anemicwomen in the cool water bath may be related to their poor thy-

roid status, since it is well known that thyroid activation is oneof the prompt physiological responses to cold exposure (24,

25). For example, acute exposure ofhumans to 4 #{176}Cair resulted

in a significant TSH elevation in 5 mm (25). Although it was

2.8 - t , , I I I not statistically significant, Martinez-Torres et al (14) observed-20 0 20 40 60 80 100 a 10% lower plasma T3 concentration in severely iron-defi-

cient-anemic men and women (average Hb concentration 75g/L) when compared with control subjects. In contrast, we ob-

served a quite pronounced and highly significant difference inT3 concentrations between the anemic and control women.This discrepancy is not easily explained; however, perhaps ouruse ofonly women with a certain body fatness and during par-

ticular days of their menstrual cycle contributed to a muchsmaller within-group variance. An additional and startling

difference between our study and that of Martinez-Torres et al(14) is the metabolic rates ofthe subjects. The anemic Venezue-lan subjects had significantly higher metabolic rates than did

the control subjects after 1 h of cool-water exposure ( 10.3 vs

5.7 mL . min� . kg�), whereas we observed a consistently lower

metabolic rate in our women, who were admittedly far less ane-

mic. We surmise that the previously discussed leanness of the

South American subjects and perhaps their much greater sever-ity of iron-deficiency anemia significantly perturbed the rela-

tionship ofiron status to thermoregulation in these subjects.

THERMOREGULATION AND IRON DEFICIENCY 817

7.2

�::: 6.8

.:�:

.� 6.0E 5.6

�5.2

E 4.8‘-, 4.4

0 4.0.> 3.6

3.2

FIG 3. V02 response ofthe 1 2 C, 10 IDA, and 8 IDP women exposedto a 28 #{176}Cwater bath both before and after iron supplementation. SeeFigure 2 for symbol representation. VO2 ofIDA was significantly lower

than C (P = 0.04) at 100 mm and than IDP at 40 (P = 0.002) and 100(P = 0.02) mm, presupplementation. There was a significant decreasein V02 ofIDP at baseline (-lO)(P = 0.01), 40 (P = 0.05), and l00(P

= 0.03) mm after 12 wk ofiron supplementation. VO2 ofIDA signifi-cantly lower than C at baseline (P = 0.02), 40 (P = 0.03), and 100 (P= 0.006) mm, postsupplementation.

groups in either the cross-sectional comparisons or the longitu-

dinal comparisons (Fig 6). Similarly, by using log-transformed

data to normalize the distributions, we found that plasma epi-

nephrine concentrations were highly variable and generally

were not different among groups (Fig 7).

Discussion

This investigation demonstrates clearly that iron-deficiency

anemia significantly alters the ability ofyoung women to main-tam body temperature when a moderate cold stress is applied.The IDA women had a significantly lower core body tempera-ture after 100 mm ofcool water exposure and did not increasetheir metabolic rate as much as did the control women. Thedata in the current study are in modest agreement with theiron-deficient-anemic group in the study of Martinez-Torres et

al (14). That is, in both studies the anemic subjects had si�nifi-

cantly lower body temperatures than did controls. In our study,

rectal temperature dropped �-0.8 #{176}Cafter 100 mm in 28 #{176}Cwater; the much more anemic Venezuelan subjects had a 0.9

#{176}Cdrop in oral temperature in the cool water at 60 mm.Whereas we observed a 64% increase above baseline in V02 at

100 mm of cold exposure, Martinez-Torres et al ( 14) observed

a 100% increase in metabolic rate at 60 mm. The drop in rectal

The nonanemic women with depleted iron stores in this

study actually thermoregulated better than did the control sub-jects when core body temperature was examined. The obvious

contrast to the nonanemic iron-depleted group in the sampleofMartinez-Torres et al (14) is confounded because ofan aver-

age 10 g/L lower hemoglobin concentration in their nonane-mic group compared with their control group. On the basis of

a conceptual model that functional sequelae of iron deficiency

occur only when stores are depleted, we would have anticipatedsome defect in temperature regulation (9, 26). The comparablemean thyroid hormone (ie, T3 and T4) concentrations and sig-

nificantly higher core temperature in this group ofwomen rela-

tive to the control group may thus reflect overcompensatory

activity to a defect in some other part of neurohormonal ormetabolic activity. This clearly is speculation, however, and re-mains to be addressed in other studies.

Published studies in both iron-deficient and hypothyroid rats

show very poor thermoregulatory capacity (8-I 2, 27). Iron-dc-

TABLE 3

Energy expended during the cool water bath5

Control(n = 12)

Anemic(n = 10)

Iron dep(n =

leted

8)

Pre Post Pre Post Pre Post

Energyexpended(kcal) 143.0± 16.3#{176} 139.8±21.5#{176} l3l.6±25.Oa 135.8±21.6#{176} lS8.3±2l.3�� 136.7 ±23.2Lc

S j; � SD. Means with different superscripts are significantly different (P < 0.05) for pre- vs post-comparisons within a group and for pre-

comparisons and post-comparisons across groups.

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

BEARD ET AL

w 13zIXIa-Li

�9IX#{176}oEZ%�� 7<�U,

� 5a-

3 I I I

TIME (minutes)

818

-J

0

E

F_re)

C’)

a-

FIG 4. Plasma T3 response ofthe 12 C, 10 IDA, and 8 IDP womenexposed to a 28 #{176}Cwater bath both before and after iron supplementa-tion. See Figure 2 for symbol representation. IDA had significantlylower plasma T3 concentrations than did C both pre- (P = 0.0001 for

all three time points) and postsupplementation (P � 0.0001 at baselineand 40 mm: P = 0.0008 at 100 mm) and than IDP at baseline (P= 0.000 1),40 (P = 0.000 1 ), and 100 (P = 0.0004) mm presupplementa-

tion and at baseline(P = 0.007)and 40(P = 0.008) mm postsupplemen-tation. With iron supplementation, IDA had a significant (P = 0.02)

increase in plasma T3 concentration at baseline and IDP had a signifi-cant decrease in plasma T3 concentration at baseline (P = 0.0 1) and 40(P = 0.05) mm. IDP had significantly lower plasma T3 concentrations

than C at baseline (P = 0.03) and 40 (P = 0.009) and 100 (P = 0.03)

mm postsupplementation.

ficient rats have a less responsive thyroid system as measuredby changes in TSH, T4 , and T3 concentrations after acute coldexposure (9, 12). The responsiveness was a direct and linearfunction ofseventy ofanemia up to a plateau point, after which

it appeared far less dependent on anemia per se. When iron-

deficient-anemic rats were provided with T3 by exogenous

means, the rats were able to thermoregulate adequately (10).

110

100

go

80

70

-J

0

EC

U)

a-

-20 #{212} 20 40 60 80 100

TIME (minutes)

FIG 6. Plasma norepinephrine response ofthe 12 C, 10 IDA, and 8IDP women exposed to a 28 #{176}Cwater bath both before and after ironsupplementation. See Figure 2 for symbol representation.

More recently, we used a less complex situation than cold stressto examine the responsivity ofthe thyroid system as a functionof iron nutriture (12). After injection of graded doses of thy-roid-releasing hormone (TRH), iron-deficient-anemic rats hada blunted and delayed TSH, T4, and T3 response; in addition,

the turnover rate ofT3 in plasma was �-50% ofnormal. Thus,while a direct causal relationship is not established in the pres-

ent study, it seems reasonable that an alteration in thyroid hor-mone metabolism plays a role in poor thermoregulation in

iron-deficiency anemia.

Animal studies have suggested a strong dependency of the

peripheral sympathetic nervous system activity on the severityof iron deficiency (8, 1 1, 13). In contrast, to the study ofMartinez-Torres et al ( 14) and some of the studies of urinary

catecholamine excretion in iron-deficient infants (28), we ob-served no consistent effect ofiron-deficiency anemia on plasma

catecholamine concentrations. Thus, in spite of clear data onchanges in norepinephrine turnover in iron-deficient rats (8,29), this study cannot provide any data that would support this

as a functional consequence in humans.The present study in humans thus provides a certain degree

3500Li

� 3000IX

� � 2500Li-i

� � 2000

� 1500

� � 2� �0 6� 8� 1�0

TIME (minutes)

FIG 7. Plasma epinephrine response of the 12 C, 10 IDA, and 8IDP women exposed to a 28 #{176}Cwater bath both before and after ironsupplementation. See Figure 2 for symbol representation. Plasma epi-nephrine values represent log-transformed data reconverted to originalunits. SEM bars have been omitted forthis variable because of difficultyof interpretation.

-20 0 20 40 60 80 100

TIME (minutes)

FIG 5. Plasma T4 response ofthe 12 C, 10 IDA, and 8 IDP women

exposed to a 28 ‘C water bath both before and after iron supplementa-tion. See Figure 2 for symbol representation. IDA had significantlylower plasma T4 concentrations than did C (P = 0.002, baseline: P

= 0.0006, 40 mm: P = 0.0005, 100 mm) and IDP (P = 0.0001 for allthree time points) presupplementation. Postsupplementation plasma

T4 concentrations of IDA were significantly lower than were C values

at baseline (P = 0.001) and at 40 and 100 mm (P = 0.002) and thanIDP at baseline and 40 mm (P = 0.003) and 100 mm (P = 0.006).

d\ ,u � � #{149}- � � � � �

:2’s:;’: :: �: � � � � � �� U

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from

THERMOREGULATION AND IRON DEFICIENCY 819

of agreement with the animal studies but does not clarify the

issue of the dependency of impaired thermoregulation in sub-

jects with anemia. The failure to completely normalize the con-

centrations ofT4 or T3 and to completely correct the poor ther-

moregulatory performance of the anemic women with I 2 wk

of 79 mg elemental Fe supplementation/d was disappointing.

We feel this is most likely related to our failure to completely

and fully replenish the iron stores ofthe anemic subjects despitethe correction oftheir anemia. Unpublished data from our lab-oratory and the data ofDillman et al(l0)demonstrate that iron

dextran treatment in rats normalizes thyroid hormone concen-trations, sympathetic nervous system activation, and thermo-

regulatory capacity within 3-7 d of injection, thus proving thereversibility ofthese defects. We found it unfeasible to require

the subjects to come to the laboratory each day to receive their

iron supplement, and we used no direct measures of compli-

ance during the supplementation phase. Thus, although com-

plete normalization offunction with repletion was not demon-

strated in the present study, animal experiments suggest thatfunction is restored with iron repletion.

This experiment was designed to examine thermoregulatoryperformance in iron deficiency as a tool for asking questionsabout the impact ofiron deficiency on energy metabolism. Be-

cause many ofthe same neurohormonal systems that alter met-abolic rate and heat production during cold stress also govern

resting metabolic rates in everyday life, these data suggest to us

that thyroid-hormone-dependent energy metabolism is sig-nificantly altered in iron-deficient-anemic humans. These data

also challenge the notion that the catecholaminergic limb ofthe metabolic machinery is altered in iron-deficient humans,

although further studies ofsubjects with a greater range of iron-deficiency anemia may prove otherwise. El

We thank Karen Glyde, Helen Kor, Steve Smith, Joe Loomis,

Douglas Johnson, Larry Barlett, and Fred Weyandt for the expert tech-

nical advice and help in conducting these studies. Most importantly,

we thank the subjects for their participation.

References

I . DeMaeyer E, Adiels-Tegman M. The prevalence ofanemia in theworld. World Health Stat Q 1985:38:302-16.

2. Expert Scientific Working Group. Summary ofa report on assess-ment ofthe iron nutritional status ofthe United States population.Am J Clin Nutr 1985:42:1318-30.

3. Dallman PR. Biochemical basis for the manifestations of iron de-

ficiency. Ann Rev Nutr 1986:6:13-40.

4. Dallman PR. Manifestations of iron deficiency. Semin Hematol1982:19:19-30.

5. Cook JD, Lynch SR. The liabilities of iron deficiency. Blood1986: 68:803-9.

6. Scrimshaw NS. Functional consequences ofiron deficiency in hu-man populations. J Nutr Sci Vitaminol (Tokyo) 1984: 30:47-63.

7. Beard JL, Finch CA. Iron deficiency. In: Clydesdale F, Weimer

KL, eds. Iron fortification of foods. New York: Academic Press,1985.

8. Beard J, Tobin B, Smith SM. Norepinephrine turnover in iron de-ficiency at three environmental temperatures. Am J Physiol

1988; 255:R90-6.

9. Beard J, Green W, Miller L, Finch C. Effect ofiron-deficiency ane-mia on hormone levels and thermoregulation during cold expo-sure. Am J Physiol l984;247:Rl 14-9.

10. Dillman E, Gale C, Green W, Johnson DG, Mackler B, Finch C.Hypothermia in iron deficiency due to altered triiodothyroninemetabolism. AmJ Physiol l980;239:R377-8l.

1 1. Dillmann E, Johnson DG, Martin J, Mackler B, Finch C. Cate-cholamine elevation in iron deficiency. Am J Physiol 1979;237:

R297-300.12. Beard J, Tobin B, Green W. Evidence for thyroid hormone defi-

ciency in iron-deficient anemic rats. J Nutr 1989; 1 19:772-8.13. Groeneveld D, Smeets HGW, Kabra PM, Dallman PR. Urinary

catecholamines in iron-deficient rats at rest and following surgical

stress. Am J Clin Nutr 1985:42:263-9.14. Martinez-Torres C, Cubeddu L, Dillmann E, et al. Effect of expo-

sure to low temperature on normal and iron-deficient subjects. Am

J Physiol 1984;246:R380-3.1 5. Akers R, Buskirk ER. An underwater weighing system utilizing

“force cube” transducers. J Appl Physiol l969;26:649-52.16. Webb P. 24-hour energy expenditure and the menstrual cycle. Am

J Clin Nutr 1986:44:614-9.17. Lu Y, Lynch SR, Cook JD, Madan N, Bayer WL. Use of capillary

blood for the evaluation of iron status. Am J Hematol 1987:24:365-74.

18. Flowers CA, Kuizon M, Beard JL, Skikne BS, Covell AM, CookJD. A serum femtin assay for prevalence studies ofiron deficiency.

AmJ Hematol l986;23:l4l-5l.19. Bioanalytical Systems, Inc. LCEC application note no. 14. West

Lafayette, IN: Bioanalytical Systems, mc, 1982.

20. McArdle WD, Katch Fl, Katch VL. Exercise physiology, energynutrition, and human performance. 2nd ed. Philadelphia: Lea andFebiger, 1986.

21. Weir JB de V. New methods of calculating metabolic rate with

special reference to protein metabolism. J Physiol (London)1949; 109:1-9.

22. Lusk G. The elements ofthe science ofnutrition. Philadelphia: WB

SaundersCo, 1928.

23. McArdle WD, Magel JR. Gergley Ti, Spina Ri, Toner MM. Ther-mal adjustment to cold-water exposure in resting men and women.J Appl Physiol 1984;56:l565-7l.

24. Hefco H, Krulich L, Illner P. Larsen PR. Effect ofacute exposure tocold on the activity ofthe hypothalamic-pituitary-thyroid system.Endocrinology 1975:97:1 185-95.

25. O’Malley BP, Cook N, Richardson A, Barnett DB, Rosenthal FD.Circulating catecholamine, thyrotrophin, thyroid hormone andprolactin responses ofnormal subjects to actue cold exposure. ClinEndocrinol l984;2l:285-9l.

26. Siimes MA, Refino C, Dallman PR. Manifestations of iron defi-ciency at various levels of dietary iron intake. Am J Clin Nutr1980; 33:570-4.

27. Sellers EA, Flattery KV, Shum A, Johnson GE. Thyroid status inrelation to catecholamines in cold and warm environments. Can JPhysiol Pharmacol 197 l;49:268-75.

28. Voorhees ML, Stuart Mi, Stockman JA, Oski FA. Iron deficiencyanemia and increased norepinephrine excretion. J Pediatr1 975;86:542-7.

29. Beard J, Tobin B. Feed efficiency and norepinephnne turnover iniron deficiency. Proc Soc Exp Biol Med 1987: 184:337-44.

by guest on January 23, 2015ajcn.nutrition.org

Dow

nloaded from