CLIN. CHEM. 33/2, 273-277 (1987)

CLINICALCHEMISTRY, Vol.33, No. 2, 1987 273

Plasma Iron and TransferrinIron-BindingCapacity Evaluatedby ColonmetncandImmunoprecipitationMethodsHelmut A. Huebers, Mary J. Eng, Betty M. Josephson, Naruemol Ekpoom, Rebecca L. Rettmer, Robert F. Labb#{233},Pensrl Pootrakul, andClement A. FInch1

We evaluated plasma iron (P1)and total iron-binding capacity(TIBC) or transferrin in normal individuals and in patients withiron imbalance. The standard colorimetric measurements ofP1and TIBC and the standard isotope-dilution measurementof TIBC were compared with an immunoprecipitation methodand also with immunoelectrophoresis of transferrin. P1 con-centrations as measured by the standard and immunoprecip-itation methods agreed closely for all individuals except thosewith saturated transferrin, where nontransferrin iron in-creased the results in the standard assay. This excess iron insaturated plasma may be derived from either free iron or iron-bearing ferritin. There were also differences in TIBC betweenthe two methods. Iron-deficient sera gave higher values fortransferrin when measured by immunoelectrophoresis. Un-saturated iron-binding capacity was increased in the isotope-dilution method in some iron-saturated plasma, compound-ing errors when added to erroneously high P1 values tocompute TIBC. Perhaps some exchange of iron occurredbetween added iron and transferrin iron in the isotope-dilution method. These measurements confirm the accuracyof the standard colorimetnc method of measunng P1 andTIBC except in iron-saturated plasma. However, the greaterspecificity of a polyclonal immunoprecipitation method ofmeasuring P1and TIBC makes it particularly useful in differ-entiating transferrin-bound iron from nontransferrin iron.

AddItIonal Keyphrases: iron-deficientanemia hemochromato-thalassemia isotope-dilutionradioassay . differentiat-

ing transferrin-boundand non-transferrin-boundiron referencevalues

hon is transported to body tissues by a protein, transfer-rin, in plasma. This protein has two high-affinity bindingsites for iron (1). Determination of total iron-binding capaci-ty (TIBC) of the plasma therefore gives a measure oftransferrmn, although it can also be directly determined.2Virtually all plasma iron (P1) normally is bound to transfer-rin, and measurement of P1 is assumed to reflect the amountof transferrin iron. The expression “transferrin saturation,”expressed as percent [(P1’TIBC) x 100], indicates the avail-ability of iron to tissues. As transferrun saturation increases,there is an increase in the amount of diferric transferrin (2),which has a greater capacity to deliver iron than doesmonofemc transferrmn (3, 4).

Measurements of P1, TIBC, and transferrin saturationhave served several purposes in clinical medicine. The P1concentration and (particularly) transferrmn saturation re-

Department of Medicine, Division of Hematology and Depart-ment of Laboratory Medicine, University of Washington, Seattle,WA 98195.

‘Correspondence should be directed to this author at: ProvidenceHospital, 500 17th Ave., Seattle, WA 98122.

2Nonstandard abbreviations: TIBC, total iron-binding capacity;P1, plasma iron; UIBC, unsaturated iron-binding capacity.

Received September 18, 1986; accepted November 24, 1986.

flect the adequacy of iron supply. A saturation of <16%indicates a deficient iron supply (5), whereas a saturation of>60% as measured on more than one occasion representsexcessive iron loading owing to increased iron absorption (6)or liver disease (7). An increased transferrmn concentrationas reflected in the TIBC indicates iron depletion if the effectsof estrogen and pregnancy are excluded (1). Other character-istic changes in both P1 and TIBC are useful in the differen-tial diagnosis of various diseases, for example, the decreasein transferrin saturations associated with a decreased trans-ferrin concentration in inflammatory states (8). Accuratemeasurement of P1 and TIBC is particularly important indetermining erythroid marrow function by ferrokinetictechniques.

In any of these clinical uses, it is essential that the P1accurately represents transferrin iron and that the TIBCrepresents the true iron-binding capacity of transferrin. Inthe present study we have assessed the validity of thestandard P1 and TIBC measurements in subjects withvarious states of iron balance by comparison with a morespecific immunoprecipitation method. We also comparedimmune methods of measuring transferrin directly withother measurements of TIBC.

Materials and Methods

Blood was obtained from 15 healthy subjects of both sexes,and from 12 subjects with iron-deficiency anemia, 16 sub-jects with thalassemia, and eight with idiopathic hemochro-matosis. Serum or heparimzed plasma, obtained with careto prevent hemolysis, was stored frozen until the variousprocedures described were performed.

The colorimetric method for P1 was that described by theInternational Committee for Standardization in Hematolo-gy (9), except that the trichloroacetic acid-treated plasmasolution was not heated to 56#{176}C(this omission did not alterresults). A 30-ruin centrifugation was adequate to obtain anoptically clear supernate, and additional centrifugation didnot change the results. We used bathophrenanthroline colorindicator for determining ferrous iron. The coefficient ofvariation (CV) for the standard P1 method was 2.4, asdetermined in 12 consecutive runs done on the same day.

TIBC was determined as recommended by the Interna-tional Committee for Standardization in Hematology (10)with use of sn-grade magnesium carbonate (Mallunckrodt,Inc., St. Louis, MO 63147) to remove excess iron, leavingtransfemn-bound iron in the supernate after centrifugation.We then measured the iron content of the supernate colori-metrically as in the P1 method. We also determined TIBC bythe isotope-dilution techniques, determining unsaturatediron-binding capacity (UIBC) from the amount of radiola-baled iron solution that remained bound to transferrin afterexcess iron was removed by adsorption onto magnesiumcarbonate. From the activity in the supennate and thespecific activity of the radiolabeled iron solution, the UIBCcould be determined and used to calculate TIBC: UIBC + P1= TIBC.

274 CLINICALCHEMISTRY, Vol. 33, No. 2, 1987

To determine transfemn-bound P1 more specifically, weprecipitate human transferrin by highly purified sheepantibody to human transferrin (raised by Kent Labora-tories, Redmond, WA 98052; we supplied the antigen andpurified the antibody). This antibody was freed of transfer-nin by column elution and of excess iron by treatment withdesferrioxamine and dialysis (accepted for publication in JLab Clin Med). We determined the precipitating capacity ofthe antibody by titration with plasma from apparentlynormal subjects after having labeled the transferrin with59Fe. In the assay we used 1.5 times the amount of antibodyrequired for complete (>95%) precipitation of radioactivity,i.e., roughly sufficient to precipitate the protein equivalentof an iron-binding capacity of 5 mg/L.

In determination of P1, we mixed 0.3 mL of plasma and0.45 mL of antibody solution and kept this in crushed ice for30 miii. We then added 2 mL of bicarbonate-saline solution(5 mL of 0.95 mol/L sodium bicarbonate in isotonic 0.15mol/L saline). The mixture was centrifuged (2000 x g, 30miii, 4#{176}C).After decanting the supernate, we added to theprecipitate 0.3 mL of a “cocktail” (0.26 mol of thioglycolicacid per liter of 0.25 moIJL acetate buffer, pH 3.3), vortex-mixed the mixture for a minute three different times, andkept it at room temperature for a total of 30 mm. From thispoint on, the procedure was as described for the standard P1method: precipitating proteins with trichloroacetic acid andreacting a color reagent with the iron in the centrifugedsupernate. The CV for the irnmunoprecipitation method was2.5% (n = 12).

We also measured the TIBC by immunoprecipitation. Inthis procedure, 0.3 mL of normal plasma is mixed with 1.5pg of 59FeSO4, pH 2, in 0.2 mL of isotonic saline, incubatedfor 15 miii at room temperature, then shaken with antibodysolution and placed on ice. After 30 miii 2.0 mL of thebicarbonate-saline solution is added, mixed, then centri-fuged (2000 x g, 30 mm, 4#{176}C).The supernate is thencompletely removed and discarded. We measured the radio-activity in the precipitate and calculated binding capacity asin other isotope-dilution procedures. After counting its ra-dioactivity, we treated the precipitate as described above forthe immunoprecipitation determination of P1.

Transferrin protein was determined by “rocket” immune-electrophoresis (11, 12) and was standardized by use ofpurified human transferrin [no. T5791; Sigma ChemicalCo., St. Louis, MO 63178, or our own preparation (13)].About 50 mg of either form of apotransferrmn was dissolvedin 5 mL of iron-free water, and the solution was centrifugedand passed over a sterile filter (0.45-pm pore size, Millex;Millipore Corp., Bedford, MA 01730) to remove traces ofTyndall-active insoluble material. The solution was thendiluted 10-fold and the absorbance measured at 280 nm. Themeasurement was converted to transferrin concentration byuse of the equation 1.09 - A = 1 g/L. The theoretical iron-binding capacity of 1.4 mg per gram of protein was verifiedby spectrophotometric titration at 465 nm for iron (14, 15).We encountered considerable difficulty in preparing a stan-dard solution having a known concentration of transfernin,owing to the presence of variable amounts of residual waterand other contaminants in the lyophilized preparations.Although we observed no change in rocket electrophoresiswith different degrees of iron-loading of transfernun, thepresence of various plasma components somewhat increasedthe apparent amount of transfernin as compared with apurified transferrmn solution.

To determine whether the iron measurements we used

were affected by contaminating hemoglobin, we made la-beled preparations of these substances, ferritun by in vitrotagging as described elsewhere (16) and hemoglobin byirjecting 59Fe into a rat and from the harvested erythrocytespreparing a hemolysate to contain 7-12 pg of iron perdeciliter (1.25-2.15 pmoJJL) (17). Tnichloroacetic acid, “iron-free,” was from GFS Chemicals, Columbus, OH 43223. Instudies of the effect of copper on P1 determination, wemeasured copper by atomic absorption (15).

Differences between results by the various methods wereevaluated by the Spearman rank correlation test.

Results

P1 values by standard and immunoprecipitation methodsare shown in the first two columns of Table 1. In normalsubjects, results by the standard colorimetnic method ex-ceeded by an average of 5% those obtained by immuno-precipitation. In the case of iron-deficient subjects, however,results by both methods were nearly identical. In partlyiron-saturated patients with thalassemia, the mean P1value by the standard method was 3% less than that by theimmunoprecipitation method. In patients with saturatedtransfemin due to thalassemia or hemochromatosis, meanvalues for P1 by the standard method were respectively 9and 11% higher than by the iinmunoprecipitation technique(p <0.01). These latter results suggest that the plasma ofpatients with saturated transferrin contained additionalchromogenic material, presumably iron, that was measuredby the standard colorimetric method but not by the immune-precipitation method.

Mean values for TIBC are shown in columns 3 through 6in Table 1. Standard and imniuno methods were eachperformed by colorimetric measurement and by the isotope-dilution technique. Results by all four determinationsagreed well for each of the normal subjects. In iron-deficientsubjects, the mean value by the standard colorimetric meth-od was 5% more than by the standard isotope-dilutionmethod, while that by the immunoprecipitation-colonimetricmethod was 2% more than by the corresponding isotope-dilution method. In thalassemic patients with unsaturatedtransfernin, values by the colonimetric methods for TIBCwere about 8% lower than by immunoprecipitation methods.In thalassemia patients whose transferrin was saturatedwith iron, the mean by the standard colorimetnic methodwas 14% less than by the standard isotope-dilution method(p <0.01), while the immune precipitation colorimetricmethod was 7% less than the corresponding isotope dilutionmethod (p <0.01). In patients with idiopathic hemochroma-tosis, results by three of the four methods agreed closely, butby the standard isotope-dilution method TIBC was 6%higher (p <0.05).

By the different methods, mean values for transfeminsaturation in normal subjects ranged from 28 to 30%; thosefor 12 iron-deficient subjects ranged from 2-7%, with meanvalues of 4-5%. Of particular interest was the oversatura-tion of transfernin values by the standard colonimetricmethod in nine of 10 thalassemic subjects with iron overloadand in six of eight patients with idiopathic hemochromato-sin. Oversaturation was also observed in five of 10 thalasse-mic patients with iron overload when the immune precipita-tion result for plasma iron was compared with the colorimet-nc immune precipitation result for TIBC. Oversaturationwas not observed in these thalassemic patients with isotope-dilution methods of measuring TIBC, and in only onepatient with idiopathic hemochromatosis.

Table 2. Effect of Added Ferrous Sulfate on Results ofColorlmetrlc and Immunopreclpitatlon Assays

Standard colorimetrlc Immune precIpitation

AddedFe2

FeIncrease TIC

FeIncrease

pg per decllfter of plasma, mean ± SD Iron saturatIon of transferrln, %

CLINICALCHEMISTRY, Vol.33, No. 2, 1987 275

Direct measurements of transferrin concentration byrocket immunoelectrophoresis are also included in the finalcolumn of Table 1. Mean values were similar to other TIBCmeasurements and the only significant difference was insubjects with iron deficiency (p <0.01).

One concern in measurements of plasma iron was wheth-er color-producing substances other than transferrin ironmay have been included in the results. One such substancewas copper. Addition of various amounts of copper, either ascopper sulfate or as ceruloplasmin, gave a color reaction thatwas read as iron in an amount equivalent to 4% of thecopper present. However, atomic absorption analysis of theimmunoprecipitate showed no detectable copper. Other sub-stances that might contribute iron to the P1 determinationinclude femitin, hemoglobin from hemolyzed blood, and freeiron. When ferritin was added to plasma, about a third of itsiron was detected by the standard colorimetric P1 determi-nation. But only 1 to 2% of the iron in ferritin was measuredin the immunoprecipitation method of determining P1.Further, in the standard colorimetric method for TIBC, halfthe radioactivity of radioiron-tagged ferritin was removedby magnesium carbonate, presumably owing to adsorptionof the ferritun itself. Results of neither the standard methodnor the immune precipitation method were affected byhemolysis, even when hemolyzed plasma containing 500 to5000 pg of hemoglobin iron per liter (9 to 90 pmol/L) wasrepeatedly frozen and thawed, then stored for a week in thefrozen state before analysis. Free iron, added as ferroussulfate to normal and transferrin-saturated plasma, wasdetected by the standard colorimetric method for iron inplasma (Table 2). Only iron bound to transferrin wasdetected by the immune precipitation method.

DIscussion

The validity of the standard methods for measuringplasma iron and TIBC has been examined by comparingresults with those by an immunoprecipitation method inwhich transferrin is selectively precipitated from plasmaand its iron content determined. The greater specificity ofthe latter method was shown by its lack of reactivity withcopper, ferritin, hemoglobin, and free plasma iron. Fornormal subjects, excellent agreement was found between

results by the two methods, and these compare well withdata from the literature. Mean values for P1 and TIBC fromindividual reports vary considerably, but if one collatesseveral reports of 10 or more subjects (18-32), the overallmean value for P1 is about 100 pg/dL (18 pmol/L) and forTIBC about 3.34 mgfL (60 pmolfL) with a transferrinsaturation of 28%. Our normal values for P1 average 96pg/dL (17 pmol/L) and for TIBC 337 1ug/dL(61 pmol/L), withtransfemmn saturation of 28%.

The clinical utility of these methods lies in their use in thedetection of abnormalities in either P1 or T1BC. Thereforewe examined individuals with iron deficiency and overload.Our results in cases of iron deficiency again are similar forthe two methods, but in iron overload differences appeared.Values for plasma iron were significantly (p <0.01) higherwith the colorimetric method than with the immunoprecip-itation method in patients whose transferrin was nearly orcompletely saturated. That the presence of nontransferriniron explains this difference seems further substantiated bythe 9% lower than expected measurement for transferrinprotein. Larger differences were reported by others in com-paring a standard colorimetric TIBC with a measurement

P1 P1 TIBC

Concn, pg/dLNormalplasma

o 106 - 298 101 - 33320 125 19 310 122 21 33460 175 69 306 169 68 343

Hemochromatosisplasma0 224 - 225 229 - 254

20 249 25 231 222 -7 25760 303 54 231 226 -3 251

Thalassemiaplasma withsaturatedtransfemn0 299 - 254 251 - 276

20 322 21 257 247 -4 27760 374 75 259 259 8 271

Table 1. Plasma Iron and BIndIng-CapacIty Measurements1

SC IPC TSC TSI TIPC TIPI TF SSC SSI SIPC SIPI

I. Normal (n = 15)Mean 99 94 334SD ±26 ±23 ±34

II. Irondeficiency(n = 12)Mean 16 17 386SD ±5 ±5 ±75

III. Thalassemia(a) unsaturated Tf (n = 6)Mean 100 103 157SD ±22 ±23 ±34(b) saturated Tf(n = 10)Mean 235 218 210SD ±58 ±55 ±46

IV. Idiopathichemochromatosis(n = 8)Mean 232 212 228SD ±19 ±23 ±17

341±37

335±19

336±21

32546

30±8

29±8

28±8

28±8

369±73

377±65

369±61

40990

4±2

4±2

4±1

5±1

160±29

168±26

174±32

17124

65±13

63±13

62±12

60±11

244±61

222±49

240±58

21143

114±15

96±1

98±12

91±2

241±16

226±27

229±24

24121

101±3

96±4

94±6

93±1

a Abbreviations: SC, standardcolonmetric method; IPC, immunoprecipitationmethod;TSC, standardcolorimetricmethod;TSI,standard isotope method; TIPC,lmmunoprecipitatlon-colonmetncmethod; TIPI, immunoprecipitation-isotope dilution method; TF, transfemn by immunoelectrophoresis; SSC, SC/TSC; SSI,FC/TSI;SIPC,IPC/TIPC; SIPI, IPC/TIPI.

276 CLINICAL CHEMISTRY, Vol. 33, No.2, 1987

based on transferrin determination (33-36). Other investi-gators have added radioiron to the plasma of patients withiron overload and have demonstrated incomplete binding(37), although this method can only demonstrate the propor-tion of added iron that does not bind to transferrin. Whilethese studies, particularly those involving the Hershkotechnique, indicate the presence of a fraction of easilychelatable and possibly free iron, the presence of iron-bearing ferritin in circulation is also to be considered (38).

The four TIBC determinations showed little differencebetween normal and iron-deficient subjects, but differenceswere observed in iron-saturated subjects. Of particularinterest was the higher TIEC with isotope-dilution tech-niques; for example, compare columns 3 and 4 and columns5 and 6 in Table 1 for patients with thalassemia andsaturated transferrin (p <0.05). That the values in columns4 and 6 are too high is suggested by the significantly lowervalues by transfemin analysis (p <0.01).

In our hands, direct measurement of TIBC by rocketimmunoelectrophoresis was less accurate than by the othermethods. The principal problem encountered, however, wasin calibration of the method, which depends on a prepara-tion of transferrin standard that demonstrates an iron-binding ratio of 1.4 or, alternatively, standardizationagainst TIBC as determined by the standard colorimetricmethod. We could not confirm reports that the immunoelec-trophoretic method is influenced by the degree of transfemmnsaturation with iron (39), but we saw a slight differencebetween results for pure transferrin and transferrin addedto plasma. Direct measurements of transfemin by rocketimmunoelectrophoresis in normal subjects were in generalagreement with TIBC measurements except for significant-ly (p <0.01) higher transferrin values in iron-deficientsubjects.

While this study has dealt with a comparison of methodsfor determining plasma iron and TIBC, other problems thanthose discussed must be kept in mind. A constant concern isthe presence of external contamination of the plasma sam-ple with iron. Fortunately, the use of disposable plasticsyringes and tubes and of aluminum needles has minimizedthis problem, but special attention to the iron content ofreagents is needed and it would seem important to keep thereagent blank to <12 pg/dL (2.2 pxnol/L). Plasma copper is aconsideration, but the 4% increase in apparent iron inplasma that may be caused by copper interference whenbathophrenanthroline is used, as in these studies, does notconstitute a serious problem. It may have been responsiblefor the 5% difference in normal subjects between the stan-dard and immunological measurements of plasma iron but,if so, it is peculiar that we did not see the same contributionfrom plasma copper in patients with iron deficiency. Ininstances where copper concentrations are markedly in-creased-as in certain patients with Wilson’s disease, asignificant error can exist.

Much greater effects on plasma iron have been observedafter injection of iron dextran, which is partly measured bythe standard method (40) and some of which was also foundto adhere to the precipitate and be measured by the immunomethod. The other cause of a very high iron in plasma is thepresence of tissue ferritin released from the liver or someother damaged tissue of high iron content. This may besuspected when isotope-dilution measurements indicate thepresence of unsaturated transferrin in the presence of aplasma iron concentration exceeding 4 mg per liter ofplasma (72 pmol/L).

This work was supported in part by NIH Grants HL06242 and 1P30 AM35816-01.

ReferencesL Morgan EH. Transferrin: biochemistry, physiology and clinicalsignificance. In: Baum H, Gergely J, eds. Molecular aspects ofmedicine. Pergamon Press Ltd, Great Britain, 1981:1-123.2. Huebers HA, Josephson B, Huebers E, Csiba E, Finch CA.Occupancy of the iron binding sites of human transferrin. Proc NatlAced Sci USA 1984;81:4326-30.3. Huebers H, Csiba E, Huebers E, Finch CA. Competitive advan-tage of diferric transfenin in delivering iron to reticulocytes. ProcNail Acad Sci USA 1983;80:300-4.4. Huebers H, Csiba E, Huebers E, Finch CA. Molecular advantageof diferric transferrin in delivering iron to reticulocytes: a compara-tive study. Proc Soc Exp Biol Med 1985;179:222-6.5. Bainton DF, Finch CA. The diagnosis of iron deficiency anemia.Am J Med 1964;37:62-70.6. Milder M, Cook JD, Stray S, Finch CA. Idiopathic hemochroma-tosis, an interim report. Medicine 1980;59:34-49.7. Hengveld P, Zuyderhoudt FMJ, Jobsis AC, et al. Some aspects ofiron metabolism during acute viral hepatitis. Hepato-Gastroenter-ology 1982;29:138-41.8. Roeser HP. Iron metabolism in inflammation and malignantdisease. Chap. 17 in: Jacobs A, Worwood M, eds. Iron in biochemis-try and medicine, II. New York: Academic Press, 1980:605-40.9. Bothwell TH, Conrad ME, Cook JD, et al. Proposed recommen-dations for measurement of serum iron in human blood. Blood1971;37:598-600.10. Bothwell TH, Conrad ME, Cook JD, et al. The measurement oftotal and unsaturated iron-binding capacity in serum. Br J Haema-tel 1978;38:281-7.11. Harboe N, Ingild A. Immunization, isolation of immunoglob-ulins, estimation of antibody titre. In: Axelsen NH, Kroll J, Week B,eds. A manual of quantitative immunoelectrophoresis. Oslo: Urn-versitetsforlaget, 1973:161-4.

12. Laurell CB. Quantitative estimation of proteins by electropho-reals in antibody-containing agarose gel. Prot Biol Fluids ProcColloq 1967;14:499-502.13. Huebers H, Csiba E, Josephson B, Huebers E, Finch CA.Interaction of human diferric transferrin with reticulocytes. ProcNatl Acad Sci USA 1981;78:621-5.14. Aisen P, Brown EB. The iron-binding function of transferrin iniron metabolism. Sem Hematol 1977;14:31-53.15. Graham G, Bates GW. Optimizing spectrophotometric titrationfor the standardization of serum unsaturated iron binding capacity.J Lab Clin Med 1976;88:477-86.16. Kim BK, Huebers H, Finch CA. The effectiveness of oral ironchelators demonstrated by a rat bioassay. Am J Heinatol, in press.17. Pippard MJ, Johnson DK, Finch CA. Hepatocyte iron kineticsin the rat explored with an iron chelator. Br J Haematol1982;52:211-24.18. Anderson BM, Gibson RS, Sabry JH. The iron and zinc status oflong-term vegetarian women. Am J Chin Nutr 1981;4:1042-8.19. Arab L, Schellenberg B, Schlierf G. Nutrition and health. Asurvey of young men and women in Heidelberg. Ann Nutr Metab1982;26:1-244.20. Cartwright GE, Wintrobe MM. Chemical, clinical and immuno-logical studies on the products of human plasma fractionation.XXXIX. The anemia of infection. Studies on the iron-bindingcapacity of serum. Anemia Infect 1949;8:86-98.21. Cook JD, Alvarado J, Gutnisky A, et al. Nutritional deficiencyand anemia in Latin America: a collaborative study. Blood1972;38:591-603.22. Cook JD, Lipschitz DA, Miles LEM, Finch CA. Serum ferritinas a measure of iron stores in normal subjects. Am J Chin Nutr197427:681-7.23. Fischer DS, Price DC. A simple serum iron method using thenew sensitive chromogen tripyridyl-S-triazine. Clin Chem1964;10:21-3 1.

CLINICALCHEMISTRY,Vol. 33, No. 2, 1987 277

24. Jackson RT, Sauberlich HE, Skala ,JH, et al. Comparison ofhemoglobin values in black and white male U.S. military person-nel. J Nutr 1983;113:165-171.25. Klein B, Lucas LB, Searcy RL. Application of Fe(ll)-5-pyridylbenzodiazepin-2-ones to the determination of serum iron and iron-binding capacity. Chin Chirn Acts 1969;26:517-23.26. Lennartason J, Bengtsson C, Hallberg L, et al. Serum iron andtransferrin saturation in women, with special reference to womenwith howtransferrin saturation. ScandJ Haematoh 1979;23:182-96.27. Laurell C. Studies on the transportation and metabolism of ironin the body. Acta Physiol Scand 1947;14:1-129.

28 Megraw A. Single tube technique for serum total iron and totaliron binding capacity. Chin Biochem 1973;6:266-73.29. Morgan EH, Carter G. Plasma iron and iron-binding capacitylevels in health and disease: with an improved method for theestimation of plasma iron concentration and total iron-bindingcapacity. Aust Anal Med 1960;9:209-13.30. Seifert J, Ring J, Steininger J, et a!. Influence of the immuneresponse on the absorption of protein from the gut. Nutr Metab1977;21:256-8.31. Takkunen H. Iron deficiency in the Finnish adult population.Scand J Haematol 1976;25:1-91.32. Williams J. The evolution of transferrin. Trends Biochem Sci1982;7:394-7.

33. Hershko C, Graham G, Bates GW, Rachmilewitz EA. Br JHaematoh 1978;40:255-63.

34. Fargion S, Cappalhini MD, Sampietro M, Fiorelli G. Non-specific iron in patients with beta-thalassaemia trait and chronicactive hepatitis. Scand J Haematoh 1981;26:161-7.35. Anuwat.anakuhchai M, Pootrakul P, Thuvasethakuh P, Wasi P.Non-transferrin phasma iron in beta-thalassaemia/hb E and haemo-globin H diseases. Scand J Haematoh 1984;32:153-8.

36. Wang WC, Ahmed N, Hanna M. Non-transferrin-bound iron inlong-term transfusion in children with congenital anemias. J Pe-diatr 1986;108:552-7.37. Batey RG, Fong LC, Shamir 5, Sherlock S. A non-transferrin-bound serum iron in idiopathic hemochromatosis. Dig Dis Sci198025:340-6.

38. Howard RB, Balfour WM. Extreme hyperferremia in twoinstances of hemochromatosis with notes on the treatment of onepatient by means of repeated venesection. J Lab Chin Med1954;43:848-59.

39. Kourilsky FM, Burtun P. Immunochemical difference betweeniron-saturated and unsaturated human transferrin. Nature (Lon-don) 1968;218:375-7.40. Cox JSG, Moss GF, Bremner I, et al. Intravenous iron-dextran:studies on unsaturated iron-binding capacity. J Cliii Pathol196821:611-5.