although, as an interim measure, the methods seem to provide a reasonable approximation.

used here

References 1. Rodbard D. Statistical quality control routine and data processing for radioimmunoasaays and immunoradiometric assays. Clin Chem 1974;20:1255-70. 2. Ekins RP. The precision profile: its use in assay design, assessment and quality control. In: Hunter WM, Come JET, eds. Immunoassays for clinical chemistry, 2nd ed. Edinburgh: Churchill Livingstone, 1983:76-105. 3. Sadler WA, Smith MH. Estimation of imprecision in inununoassay quality assessment programmes. Ann Clin Biochem
1987;24:98-102.

tion of imprecision profiles, with reference to immunoassay data. Clin Chem 1988;34:1058-61. 5. Raab GM. Validity tests in the statistical analysis of immunoassay data. Op. cit. (ref. 2):614-23. 6. Raab GM, McKenzie 1GM. A modular computer program for

4. Sadler WA, Smith MH, Legge HM. A method for direct estima-

processing immunoassay data. In: Wilson DW, Gaskell SJ, Kemp KW, eds. Quality control in clinical endocrinology. Cardiff: Alpha Omega, 1979:225-36. 7. Baxter RC. Simplified approach to confidence limits in radioimmunoassay. Clin Chem 1980;26:763-5. 8. Raggatt PR. Duplicates or singletons-an analysis of the need for replication in immunoassay and a computer program to calculate the distribution of outliers, error rate and the precision profile from assay duplicates. Ann Clin Biochem 1989;26:26-37. 9. Bard Y. Nonlinear parameter estimation. New York: Academic Press, 1974:205 pp.

CLIN. CHEM. 36/7, 1350-1355 (1990)

Factors Affecting the Concentrations of Ferritin in Serum in a Healthy Australian Population

B. A. Leggett,1
N. N.

Brown,2S. J. Bryant,2

L Duplock,1

L. W.

Powell,1 and J. W. Halllday13

Reported values have varied from geometric means of 67 ugJL for men and 56 pgfL for women (3) to means of 123 gJL for men and 56 g/L for women (4). Other studies have found a range of values between those extremes (5-9).

We measured by different techniques the ferntin concentration in serum in two large asymptomatic Australian population samples: 1367 bank employees and 601 insurance corporation employees. Ethanol intake, diet, the frequency of blood donation, smoking and exercise habits, and past medical history were documented. The median concentration of ferritin in serum varied according to age and sex, but was generally higher than in previously reported populations under age 65 years. Results for the two population samples were in close agreement. Apart from the blood donation status, the most important factors influencing the concentration of ferritin in serum were ethanol intake in men and diet in women. Heavy ethanol intake was associated with increased values, even among men without evidence of liver disease. We conclude that the reference range for ferritin concentration in serum in the Australian population should be significantly increased and should be relatedto age as well as sex. This study emphasizes the need to determine local reference ranges for ferritin concentrations in serum.

Based on these studies, often-quoted reference ranges are 10-250 jgfL for males and 10-150 p.gfL for females. The determination of an accurate reference range for the ferritin concentration in serum is of particular importance in populations where the iron-overload disease, hemochromatosis, is common. In this disease the gradual accumulation of excess iron, which depends on dietary and other environmental factors, increases with age and can be monitored by measuring the concentration of ferritin in
serum.

AddItIonal Keyphrases: variation,

sexand age-related effects

source of immunoradiomefric

ethanol assay

Here we report the results of a large population survey. These results indicate that the median concentrations of ferritin in serum are markedly higher in an asymptomatic Australian population than in any previously reported populations younger than 65 years. This large survey also allowed us to examine the influence of many of the factors postulated to contribute to the considerable variations of ferritin concentrations in serum. Materials and Methods

reference range

hemochromatosis

The concentration of ferritin in serum is widely used clinically as an index of body iron stores, but correct interpretation relies on the availability of an appropriate reference range and knowledge of the variables other than iron status that affect this range. Although there is good evidence that a ferritin concentration <10 jzg/L reflects iron deficiency (1, 2), the mean and upper limits of the reference range for healthy subjects are less well defined.

1 Department of Medicine, University of Queensland, Royal Brisbane Hospital, and 2Department of Pathology, Royal Brisbane Hospital, Brisbane, Queensland 4029, Australia. Author for correspondence. Received December 22, 1989; accepted May 16, 1990.

Subjects Employees of two large Australian corporations, a bank and an insurance company, were surveyed. Volunteers gave written informed consent and completed a comprehensive questionnaire on details of smoking habits, alcohol (ethanol) intake, past medical history, exercise habits, blood donation status, and intake of both prescribed medications and other drugs, vitamins, and iron supplements. Information regarding diet, in particular meat consumption, was also recorded. A sample of venous blood was drawn from nonfasting subjects between 0900 and 1100 hours. The protocol and study were approved by the Research Ethics Committee of the Royal Brisbane Hospital and had the full cooperation of the medical advisors and management of the employing institutions.

1350

CLINICAL CHEMISTRY, Vol.36, No. 7, 1990

Methods
TOO

SANK

POPULATION

We measured the concentration of ferritin in serum by a two-site immunoradiometric assay, using either a commercial kit (Ciba Corning, Corning, NY) for the bank employees or the method of Halliday et al. (10) for the insurance company population. A complete blood count was performed with a Coulter Counter (Coulter Electronics, Hialeah, FL) and multiple biochemical analysis with a SMAC II (Technicon Instrument Corp., Tarrytown, NY). C-reactive protein was measured by immunoturbidimetry. Statistical analysis was done with the Statistical Package for the Social Sciences X. The significance of association between two variables was assessed by a chi-square test, with P <0.00 1 being regarded as significant. Analysis of variance was used to assess the correlation of the concentration (expressed as a logarithm) of ferritin in serum with multiple different variables, and multiple classification analysis was used to calculate multiple r2, to estimate the strength of the correlation.

MALES
FEMALES

#{149}00

30

I,

SERUM FE RE IT IN

400

(t0LI
300

201

Co

L
,,,.

ii..Li t_I_P1!T
303N AGE 30 (05*551 35 40AN INSURANCE

POPULATION

10

Results
We studied 1968 volunteers, 1367 from the banking corporation and 601 from the insurance corporation. The male-to-female ratio was 1.13, and the mean age was 31 years (range 17-65 years). Most subjects (82.4%) were employed in clerical positions, with 11.9% more in managerial positions and the remainder in various miscellaneous occupations; 96.6% were Caucasian, 1.8% Australian Aboriginal, and 1.6% Asian. There were no significant differences in the above characteristics between the bank and insurance populations. Figure 1 shows the distribution of ferritin concentrations in serum according to age and sex. Because the distribution was skewed, we list the results as medians and 5th to 95th percentiles. The median ferritin concentration in serum in young men increased to a plateau in the fourth decade and then varied little before the age of 65 years. In women the median ferritin concentration rose slowly until the fifth decade, when the rate of increase accelerated; however, their median concentrations never reached those of males. Both sexes showed a wide spread of values, which increased with age. Iron deficiency was most prevalent among women
younger than 49 years. The overall incidence of iron deficiency (ferritin 10 zgfL) was 2.6% for men and 8.9% for women. The influence of various lifestyle factors on the ferritin concentrations in serum was studied further in the larger (bank) population. Blood was donated three or more times per year by 15.6% of the population; the ferritin concentration showed an inverse relationship to the frequency of blood donation (Figure 2) (chi-square = 161.35, df = 20, P <0.0001). This relationship remained significant for each sex considered separately. The amount of meat consumed in the diet was also related to the ferritin concentration in serum, with vegetarians having lower concentrations (Figure 2) (chi-square = 54.81, df = 20, P <0.0001). The meat consumption of the population was high: 32.6% reported eating meat everyday, 48.6% most days, 14.5% once or twice weekly, 3.5% occasionally, and only 0.5% never. No relationship was apparent between the reported intake of iron tablets and tonics and the ferritin concentration in

50

500

$
/

(SUM (1551115 (POlL)

Co

/ / 11
200

/ / / /
/

I..I.
.

01

Fig. 1. Median concentrations of ferritin in serum according to age and sex The horizontal bais representthe median concentrationsof serum ferritin and the striped and stippled bars show the range from the 5th to the 95th percentiles in men and women, respectively. For this figure the bank population comprised 724 men and 643 women, and the insurance population 305 men and 277 women

between smoking habit and ferritin concentration. Increased concentrations of C-reactive protein in serum were not correlated with an increase in serum ferritin; however, because this was an asymptomatic, presumably healthy population, only nine subjects had a markedly increased concentration of C-reactive protein (>10 mg/L), and no subjects had values >15 mg/L. There was no relationship between report of recent illness and the concentration of
ferritin in serum. The ferritin concentration increased with increasing frequency of ethanol intake (chi-square = 157.50, dl = 20, P <0.0001). Similarly, ferritin was significantly related to the daily consumption of ethanol expressed in grams (chisquare = 13461, df = 20, P <0.0001). Moreover, the ferritin concentration was positively related to the amount

serum.
We saw no significant relationship between the frequency of exercise (described as active exercise causing sweating and breathlessness) and ferritin concentration or

of ethanol consumed in the preceding 24 h (chi-square = 139.93, df = 24, P <0.0001) and to the type of ethanol consumed (chi-square = 203.93, df = 12, P <0.0001): concentrations of ferritin were higher among subjects who drank beer rather than wine or spirits. Subjects with liver disease, as evidenced by increased concentrations in serum of enzymes of liver origin [y-glutamyltransferase (GOT) and aspartate aminotransferase], had higher ferritin concentrations (Table 1). However, the influence of ethanol CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990 1351

ISO

MEDIAN 100 SERUM FEERITIN CONCENTRATION

(pgiU 50

MEDIAN
SERUM

FERRITIN

(p510

Men,

Occaicoal

li/y..,

2o1y..,

35/yea;

131131Cr

E,q,y GCy

Uo,I
dAYS

1-2*1 week OF
MEAT INTAKE

Oec.S,on.I

N,.R,

FREQUENCY OF BLOOD DONATION

FREOUENCY

Fig. 2.
Both

(left) and meat intake (right) on median concentration associations are significant at P <0.0001 (chi-square)
Effect of blood donation

of ferntin

in

serum in 1367 bank employees

Table 1. RelationshIp of Liver-Function Test Results to ConcentratIons of Ferrltln in Serum

Ferritln concn, ig/L <10 11-100 611 (47) 12 101-250 388 251-500 >500

300

Oar <50 U/L >50 U/L

Aspartate

87
(7)

166
(13) 28 (36)

2 (3) 88
(7)

(30) 22
(29)

38 (3) 13
(17) 45

MEDIAN SERUM FE RRIT IN CONCENTRATION

(1aglIJ

(15) 621 (46) 2

(6)

aminotransferase0

<40 U/L >4OUIL

1 (3)
significant

396 (30) 14
(46)
= 4,

186
(14)

8
(26)

(3) 6
(19)

100

Differences significant atchi-square 129.46, di =

Results are shown as
Differences

P <0.0001. no. (and %) of subjects in each group. at chi-square = 37.38, df 4, P <0.0001.

was not mediated solely through induction of overt liver disease: the daily consumption of ethanol was still positively related to the concentration of ferritin in serum among men with normal concentrations of serum GOT (Figure 3) (chi-square = 29.66, df = 9, P <0.001). We performed analysis of variance both to determine whether the influence of factors such as ethanol intake and diet was still significant, once confounding factors such as age and sex were accounted for, and to measure the strength of association of the above factors with the concentration of ferritin in serum. The F-statistic (and accompanying P-value) indicated whether the factor had an independent association with the ferritin concentration in serum after consideration of previously entered variables. The multiple r2 value gave an estimate of the strength of the correlation. Sex alone was found to account for 25% of the variability in ferritin concentrations in serum (F = 438.088, df= 1,P <0.001; multiple r2 = 0.250). Because of the strength of this effect and because it was more marked among subjects in their fourth decade than at the extremes of age, we considered results for men and women separately in all further analyses. Age accounted for 7.5% of the variability of serum ferritin concentration in men and 4.1%
1352

0 DAILY

10

20-50 INTAKE (9)

60-100

ALcoHol.

Fig. 3. Effect of ethanol intake on the median concentration of ferritin in serum in 545 men with normal concentrations of GOT in serum The association Is significant at P <0.0001 (chi-square)

in women (Tables 2 and 3). The frequency of blood donation had a very marked effect, being the next most important influence on ferritin concentration after sex. In the subsequent analysis we took into account the subjects age and frequency of blood donation and examined the additional influence of various other factors. Among men (Table 2) another important factor demonstrated to affect the concentration of ferritin in serum was the amount of ethanol taken daily, which accounted for 7.3% of the variability. Other measures related to ethanol intake (frequency of intake and intake in the previous 24 h) were less strongly correlated. Once the effects of age and sex were accounted for, the type of ethanol ingested no longer had a significant effect on the concentration ol ferritin in serum. The concentration of GOT was correlated with the ferritin concentration in serum, although not as strongly as was the daily ethanol intake.

CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990

Table 2. ContributIon of Various Factors to the Variability of Ferritin Concentration In Sera from Men
F-value
df

Age, years
Frequency of blood donation, no/yr

29.080 62.052

4 5

P <0.001

ChangebIn r2

0.075
0.297

<0.001

0.372

Additional contribution after above factors are considered Daily ethanol intake, g 7.541 5 <0.001 0.073 Frequency of ethanol 5.792 5 <0.001 0.025 intake, no/week 3.980 5 0.001 0.010 Ethanol intake in preceeding 24 h, g Type of ethanol 4.514 3 0.004 (beer/wine/spirits) 9.697 3 <0.001 0.027 Serum GGT, U/L Serum AST, U/L 2.171 2 0.115 Meat intake, no./week 5 0.842 0.410 Smoking habit 1 0.658 0.196 Frequency of active 3.662 4 0.006
exercise,
Analysis
b

in serum, we calculated multiple r2 for a combination of age, frequency of blood donation, meat intake, and the concentrations of GOT in serum: it was 0.426 for men and 0.277 for women. Daily ethanol intake still had a significant association with the concentration of ferritin in serum, even after taking into account the contribution of serum GOT (Figure 3). Adding ethanol intake to the model improved the multiple r2 values to 0.463 and 0.284, respectively. Discussion The predominant influences on the concentration of ferritin in serum identified in this population were sex, blood donation, and age. Values tended to increase in all subjects with age, although the rate of increase was lower among women and regular blood donors. The increase in ferritin concentration paralleled expected physiological changes. The concentrations in women remained low until after menopause, whereas those in men reached a plateau in the fourth decade, when iron stores would be expected to be stabilized after the period of rapid growth in the second decade. This suggests that, as concluded in previous studies (7, 11), body iron stores are the most important factor determining the ferritin concentration in serum in this population; moreover, the unusually high median concentration of ferritin in serum truly reflects higher iron stores than in previously reported populations (3-9,12). Although in the two populations examined the ferritin concentrations were measured by different assays, the results were in close agreement, indicating that the high values obtained were not the result of inaccurate technique. If one assumes that 1 ng of ferritin per milliliter of serum represents 8mg of storage iron (2, 11), then the men older than 30 years in our population had approximately 1920
mg of storage iron and women in their fourth decade had 504 mg. These estimates are more than double the previously reported values of 767 and 232 mg, respectively (11).

no./week
of variance. classification analysis.

Multiple r2 calculated by using multiple AST, aspartate aminotransferase.

Table 3. Contribution of Various Factors to the Variability of Ferrltin Concentration In Sera from Women
F-value
df P

Chang& In f2 0.04 1 0.147

Age, years
Frequency

0.188 Additional contribution after above factors are considered Daily ethanol intake, g 4.981 4 0.001 0.006 Frequency of ethanol 2.696 5 0.020 -

blood donation, no./yr

of

7.080 21 .762

4 5

<0.001 <0.001

intake, no./week Ethanol intake in preceeding 24 h, g Type of ethanol (beer/wine/spirits) Serum GGT, U/L Serum AST, U/L Meat intake, no/week Smoking habit Frequency of active exercise, no/week Footnotes in Table 2. as

2.019
5.425 7.176
14.295

5 3

0.074

0.001
<0.001 <0.00 1 <0.001 0.717 0.077

0.034
0.029

3
2

0.038
0.062
-

8.531 0.131 2.118

5
1 4

___________________________

Women responded somewhat differently (Table 3). The stated ethanol intake had a far smaller effect, although variations in GOT concentrations were associated with 2.9% of the variability in ferritin concentrations in serum, a percentage very similar to that in men. Interestingly, the concentration of aspartate aminotransferase in serum was a significant factor among women but not men. The effect of meat intake also differed between the sexes, accounting for 6.2% of the variability in women but not significantly related to the serum ferritin concentration in men. As predicted by chi-square analysis, smoking habit and frequency of active exercise were not significantly related to ferritin concentrations in serum. To determine the best model to account for variability in ferritinconcentrations

There are several possible reasons for the comparatively high iron stores in this population. The ferritin concentration was higher in sera from nonvegetarians than in vegetarians, and the vast majority of the population reported eating meat on most or all days. Thus the traditional high consumption of meat by Australians probably provides a dietary iron content with a very high bioavailability. Meat consumption in the United Kingdom and Sweden, where the median ferritin concentration in serum is considerably lower, is only 68% (United Kingdom) and 54% (Sweden) of that in Australia (12). Meat consumption in the United States and Canada resembles that of the Australian population (12), although median ferritin concentrations in those countries are lower than in the present study. This suggests that factors other than iron intake are also important. However, we find it interesting that the population
with concentrations of serum ferritin most closely approximating the present one was an American one (9). A second important factor probably relevant to the high iron stores is that the present population is of relatively high socioeconomic status, with access to good health care, which reduces the incidence of diseases that lead to iron loss. Finally, in this Caucasian population, about 1 in 10 individuals may be heterozygous for genetic hemochromatosis

(13) and thus have higher rates of iron absorption. The relatively low incidence of iron deficiency in this population, as determined by a ferritin concentration <10 u.gfL, provides further evidence of the overall high iron stores. Nevertheless, 8.9% of the women were iron-defiCLINICAL CHEMISTRY, Vol. 36, No.
7, 1990 1353

cient. In other studies, however, more than 20% of the women were iron-deficient (8, 14, 15). As previously observed, we saw a wide spread of values for serum ferritin in the present study, even within groups of the same age and sex. Factors postulated to alter the concentration of ferritin in serum include exercise (16), smoking (4, 17), liver disease (3, 18, 19), and diet. There was no evidence in this population that the variability was due to the influence of smoking, exercise, or recent illness. The major factors identified apart from the frequency of blood donation were the meat intake of women and the ethanol intake of men. Diet may have had a detectable effect only in the women because they must ordinarily absorb a greater proportion of the iron in their diet to maintain iron balance (20). Therefore, a decrease in the amount of bioavailable iron in their diet by a reduction in meat intake is more likely to diminish their ability to compensate for iron losses by increased absorption. Although the ferritin concentration is known to increase in serum for many patients with liver disease (2, 18, 19), it is unclear whether moderate-to-heavy ethanol intake in asymptomatic subjects with normal liver-function tests is also associated with an increase of serum ferritin. Studies of patients with alcoholism have demonstrated that high concentrations of GOT in serum are correlated with a high ferritin concentration (21,22). The authors of those studies
suggest that a high concentration of ferritin in serum is very unlikely to be related to ethanol intake if the GOT concentration is normal. However, they studied only a few patients with normal concentrations of GOT and did not compare these patients with controls to determine if there

important consequences with regard to the use of measuring ferritin in serum to detect pathological iron overload. We suggest that the upper limit of the reference range for men older than 30 years should be greater than the recommended 300 pg/L (25), which results in an unacceptable number of false positives in this population. In view ol the association between increased concentrations of ferritin

in serum and the ingestion of 20 g or more of ethanol daily, it may be appropriate to use a reference range that excludes the results for subjects who drink more than this. However, such a reference range would still be higher than that previously recommended. This study emphasizes the need to base reference ranges for ferritin concentrations in serum on studies of local populations and to use age- as well as sex-related values. The generally high iron stores we observed may account in part for the high incidence of expression of hemochromatosis in homozygous subjects (26) in the Australian population.
This project was supported by the National Health and Medical Research Councilof Australia. We thank the staff of the Commonwealth Bank of Australia and the Australian Mutual Provident Society for their participation in this study and Mrs. Elizabeth Axelsen and Mrs. Sonja Webb for many hours of tireless work collecting and processing blood samples. References 1. Jacobs A, Miller F, Worwood M, Beamish MR, Wardrop CA. Ferritin in the serum of normal subjects and patients with iron deficiency and iron overload. Br Med J 1972;iv:206.-8. 2. Lipschitz DA, Cook JD, Finch CA. A clinical evaluation of serum ferritin as an index of iron stores. N Engl J Med 1974;290:1213-6. 3. Milman N, Pedersen NS, Visfeldt J. Serum ferritin in healthy Danes: relation to marrow haemosiderin iron stores. Dan Med Bull 1983;30:115.-120. 4. Jacobs A, Worwood M. Ferritin in serum. Clinical and biochemical implications. N EngI J Med 1975;292:951-6. 5. Rodger RSC, Fletcher K, Fail BJ, Rohman H, Sviland L, Hamilton PJ. Factors influencing haematological measurements in healthy adults. J Chron Dis 1987;40:943-7. 6. Valberg LS, Scorbie J, Ludwig J, Pelletier 0. Serum ferritin and the iron status of Canadians. Can Med AssocJ 1976;114:41721. 7. Cook JD, Lipschitz DA, Miles LEM, Finch CA. Serum ferritin as a measure of iron stores in normal subjects. Am J Clin Nutr 1974;27:681-7. 8. Cook JD, Finch CA, Smith NJ. Evaluation of the iron status of a population. Blood 1976;48:449-55. 9. Finch CA, Cook JD, Labb#{233} Culala M. Effect of blood RF, donation on iron stores as evaluated by serum ferritin. Blood 1977;50:441-7. 10. Halliday JW, Gera KL, Powell LW. Solid phase radioimmunoassay for serum ferritin. Clin Chim Acts 1975;58:207-14. 11. Walters BO, Miller FM, Worwood M. Serum ferritin concen tration and iron stores in normal subjects. J Clin Patho 1973;26:770-2. 12. Food consumption statistics 1976-1985. Paris: Organizatio for Economic Cooperation and Development, 1988. 13. Dadone MM, Kushner JP, Edwards CQ, Bishop DT, Skolnic MH. Hereditary haemochromatosis: analysis of laboratory expres sion of the disease by genotyping 18 pedigrees. Am J Clin Patho 1982;78:196-207. 14. MacPhail AP, Bothwell TH, Torrance JD, et al. Iron nutritio in Indian women at different ages. S Afr Med J 1981;59:939-42. 15. Cook JD, Finch CA. Assessing iron status of a population. J Clin Nutr 1979;32:2115-9.
16.

was some increase of ferritin in this group also. An increased GOT was associated with an increased ferritin concentration in the present study but an effect of ethanol intake was apparent even among men with normal concentrations of GOT in serum. Ethanol may have a specific effect on ferritin synthesis, or increases of the ferritin concentration in serum may be a very sensitive indicator of the effect of ethanol on the liver. The apparent lack of influence of ethanol intake on the concentration of ferritin in serum of women may be due to either a true sex-related difference or the considerably lower prevalence of heavy drinking among women. Liver disease as manifested by increased concentrations of GOT or aspartate aminotransferase was associated with high concentrations of ferritin in serum from women. Ethanol intake in the Australian population is generally high, being approximately twice that of the Swedish population (12); moreover, only 9% of American subjects drink 20 g or more of ethanol daily, compared with 32% of the Australian population studied (23). Ethanol intake in the United Kingdom is similar to that in the Australian population (24) but its effect may be offset by the lower consumption of meat in the British population (12). Not
surprisingly, therefore, the effect of ethanol on the concentration of ferritin in serum is clearer in the present study of the Australian population than in studies of populations with a much lower ethanol intake. It contributes to the higher median ferritin concentrations observed in men, but is not the only factor involved. Even in men drinking <20 g of ethanol daily, the median concentration of ferritin in serum is 195 p.gfL in the fourth decade, 165 g/L in the fifth, and 205 4ugfL in subjects of ages 50 to 65 years. This Australian population apparently has higher iron stores than any population yet reported, a finding that has 1354

Dufaux B, Hoederath A, Streitberger I, Hollman W, Assm

G. Serum ferritin, transferrin, haptoglobin and iron in middle- an long-distance runners, elite rowers and professional racing cy clists. Int J Sports Med 1981;2:43-6. 17. Touitow Y, Proust J, Carayon A, et al. Plasma ferritin in ol

CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990

age. Influence of biologicalnd pathological a factors a large in elderly population. Clin Chim Acts 1985;149:37-45. 18. Prieto J, Barry M, Sherlock S. Serum ferritin in patients with iron overload and acute and chronic liver diseases. Gastroenterology 1975;68:525-33. 19. Valberg LS, Ghennt CN, Lloyd DA, Frei JV, Chamberlain MJ. Diagnostic efficacy of tests for the detection of iron overload in chronic liver disease. Can Med Assoc J 1978;119:229-36. 20. Bothwell TH, Charlton RW, Cooke JD, Finch CA. Iron metabolism in man. Oxford: Blackwell Scientific Publications, 1979. 21. Meyer TE, Kassianides C, Bothwell TH, Green A. Effects of heavy alcohol consumption on serum ferritin concentrations. S Afr Med J 1984;66:573-5.

22. Lundin L, Hallgren R, Birgegard G, Wide L. Serum ferritin in

alcoholics and the relation to liver damage, iron state and eryth-

ropoietic activity. Acts Med Scand 1981;209:327-31. 23. The Surgeon Generals Report on Nutrition DHHS (PHS) Publication No. 88-50210. Washington

and Health,
DC: US Dept

of Health and Human Services, 1988. 24. Cade JE, Barker DJP, Margetts BM, Morris JA. Diet and inequalities in health in three English towns. Br Med J 1988;296:1359-62. 25. Worwood M. Serum ferritin. Clin Sci 1986;70:215-20. 26. Bassett ML, Halliday JW, Bryant 5, Dent 0, Powell LW. Screening for hemochromatosis. Ann NY Acad Sci 1988;526:27489.

CLIN. CHEM. 36/7, 1355-1360

(1990)

Two New Two-Step lmmunoassays for Free Thyroxin Evaluated: Solid-Phase Radioimmunoassay and Time-Resolved Fluoroimmunoassay
PIne Nuutjla,1 Perttl Kesklnen,2 Kerttu IrjaIa,2 Llnnea Llnke,2 Ifanna-Leena Kalhola,2J. U. Eskola,4 Rlsto Erkkola,3 Penttl Sepp&a, and Jorma VIIkarI1 We measured concentrations by using two new two-step
of free thyroxin (FT4) in serum FT4 assays-a solid-phase twostep radioimmunoassay, Spectna#{174}, a time-resolved and fluoroimmunoassay, Delfia#{174}-and compared the results with those by a two-step FT4 assay (RIA-gnost#{174}), one-step FT4 a

umented

analog assay (Amerlex-M#{174}), FT4 measured after equiand librium dialysis. The new FT4 assays classified 30 hypothyroid and 43 hyperthyroid patients (untreated) well. In 138 patients with nonthyroidal illness (NTI) and in late pregnancy (n = 36), fewer subnormal FT4 values were reported by Spectria (P <0.001), Delfia (P <0.001), and RIA-gnost (P <0.01) than by Amerlex-M. The results of the Spectna and Delfia methods correlated with the results of the dialysis method (r = 0.76) in NTI patients and pregnancy, and were in better agreement with the clinical state than was FT4 by Amerlex-M. The FT4 values by Amerlex-M, but not by other methods, correlated with albumin concentration. We conclude that these new two-step methods present good alternatives for FT4 analysis. AdditIonal
pregnancy
Keyphrases:

to give spurious results, as librium dialysis, in patients with (NTI) (3, 4). Our aim in this study newly introduced two-step (so-called assays in clinical practice, particularly and in patients with NTI.

compared with equinonthyroidal illness was to evaluate two

back-titration) in pregnant

Fr4
women

Materials and Methods

Subjects
Each subjects thyroid status was established on the basis of clinical evaluation and measurement of thyrotropin (thyroid-stimulating hormone, TSH) and, if necessary, the TSH response after administration of thyroliberin (thyrotropinreleasing hormone). Reference ranges of FF4 assays were established from measurements from 65 euthyroid subjects (33 men), mean age 34 (SD 9) years (range, 19 to 48 years). The diagnoses of hypothyroidism (n = 30) and hyperthyroidism (n = 43) were also influenced by Fl4 values measured by Amerlex-M#{174} ssay, because this result was availa able when the subjects were evaluated. None of the hyperthyroid patients had triiodothyronine thyrotoxicosis. Pregnant women (n 36) in the third trimester attending an antenatal clinic and inpatients with NTI (n = 138) were studied. Five groups were formed from NTI patients: group 1, patients with acute myocardial infarction (n = 16) or cerebral infarction (n = 15); group 2, patients with severe infections treated in the intensive care unit (n = 16); group 3, patients with acute leukemia (n = 15) or colorectal cancer (n = 14); group 4, patients with severe hepatic cirrhosis (n = 15) or chronic hemodialyzed renal failure (n = 13); and group 5, patients with epilepsy who were receiving phenytoin, carbamazepine, or both (n = 34). Blood was drawn in the morning. Serum was separated and kept at -20 #{176}C assayed. until FT4 Assays The two new immunoassay kits for FT4 in serum we evaluated were the Spectria#{174}r4 two-stage radioimmunoF assay (Farmos Diagnostica, Turku, Finland) and Delfia#{174}

albumin

thyroid status nonthyroidal illness intermethod comparison

Measurements

of the free thyroxin

(FF4) concentration

serum are clinically important, particularly in borderme thyroid disease and in conditions involving abnormal protein concentrations or protein binding of thyroxin (1,
)5 Of the several FT4 radioimmunoassays ible commercially, the one-step analog

currently

avail-

rained relatively

wide acceptance.

methods have However, they are dcc-

Department of Internal Medicine, 2 Central Laboratory, and Department of Obstetrics and Gynecology, University Central lospital of Turku, SF-20520 Turku, Finland. 4Wallac Biochemical Laboratory, Turku, Finland. Nonstandard abbreviations: FT4, free thyroxin; T4, thyroxin; SH, thyrotropin (thyroid-stimulating hormone); and NTI, onthyroidal illness. Received January 22, 1990; accepted May 18, 1990.

CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990

1355