Interlaboratory proficiency, intermethod comparison, and calibrator suitability in assay of serum aspartate aminotransferase activity.

Sources of variation in assays of aspartate aminotransferase (EC 2.6.1.1) activity were examined in an interlaboratory survey and through an examination of materials used as calibration materials in these assays. Four highly stable lyophilized specimens containing human cytoplasmic enzyme, with activities of 0, 22, 46, and 96 U/liter at 30 degrees C and optimal substrate concentrations, were assayed by 319 laboratories. Mean values obtained on these specimens by laboratories using 2,4-dinitrophenylhydrazine kits varied among manufacturers and deviated from values expected from this procedure. The average coefficient of variation (CV) with these kits was greater than 20%. Automated continuous-flow procedures with use of diazonium salt showed the best precision (av CV, less than 10%). However, the automated continuous-flow malate dehydrogenase/NADH coupled method produced an average CV greater than 20%. Results from each of the automated methods were related to a reference malate dehydrogenase/NADH coupled continuous kinetic assay method by temperature relationships alone. Mean values from manual diazonium salt procedures were 1.7-fold greater than similar reference values (av CV was 18%). The higher results were attributed to the use of poorly-defined units and to an artifact caused by chromophore stabilizers in this procedure when aqueous samples are used. The average CV in continuous kinetic methods varied among kit manufacturers, ranging from 6 to 28% for the specimen of highest activity. Variations in results were much larger at 366 nm than at 340 nm than at 340ity. Variations in results were much larger at 366 nm than at 340 nm. Interassay relationships of these methods are presented. Concentrations of pyruvate in commercially available calibration materials differed between manufacturers, varied in stability, and deviated from the expected concentration. For some colorimetric assays the precision attained on reported absorbance values for the enzyme specimens was of the same order of magnitude as that for pyruvate standards. Other sources of error are revealed by the interlaboratory survey. The value of commercially available sources of enzyme activity as calibration or control materials was assessed by evaluating the following properties: activity at suboptimal concentrations of L-aspartate or 2-oxoglutarate, temperature effects, preincubation lability owing to aspartate and phosphate, pyridoxal phosphate saturation, contamination with glutamate dehydrogenase, and manufacturer's rated activity. These properties are compared to those of human cytoplasmic enzyme in a human serum matrix.     

Mitochondrial enzymes in human serum: comparative determinations of glutamate dehydrogenase and mitochondrial aspartate aminotransferase in healthy persons and patients with chronic liver diseases

We measured the activities of two mitochondrial enzymes, the mitochondrial form of aspartate aminotransferase (EC 2.6.1.1) and glutamate dehydrogenase (EC 1.4.1.2), in the serum of apparently healthy persons (n = 84) and patients suffering from chronic liver diseases (n = 43). The distribution of activities for glutamate dehydrogenase, but not mitochondrial aspartate aminotransferase, was sex-dependent. The upper limits of the reference intervals (99th percentile) at 37 degrees C were 3.2 U/L for mitochondrial aspartate aminotransferase, 6.4 U/L for glutamate dehydrogenase (women), and 11.0 U/L for glutamate dehydrogenase (men); there was a weak correlation between the activities of both mitochondrial enzymes (r = 0.439). In patients with chronic liver diseases we found a greater increase in the activity of glutamate dehydrogenase than of mitochondrial aspartate aminotransferase and the correlation between the two mitochondrial enzymes was stronger. The diagnostic sensitivity and specificity of either mitochondrial enzyme was less than that of total aspartate aminotransferase, alanine aminotransferase (EC 2.6.1.2), or gamma-glutamyltransferase (EC 2.3.2.2).     

Changes in serum enzyme activity after transcatheter arterial embolization for hepatic neoplasm

We assayed serum levels of certain enzymes and tumor markers in patients after transcatheter arterial embolization (TAE) to evaluate the effectiveness of this treatment. Twenty patients had hepatocellular carcinoma and two patients had metastases to the liver from colon cancer. Assays were first done immediately after TAE and were continued for the next 12 days. Glutamic oxaloacetic transminase (GOT; EC 2.6.1.1, -aspartate:2-oxoglutarate aminotransferase), glutamic pyruvic transaminase (GPT; EC 2.6.1.2, -alanine:2-oxoglutarate aminotransferase), and lactate dehydrogenase (EC 1.1.1.27; (S)-lactate:NAD⁺ oxidoreductase) peaked 24 to 48 h after TAE and returned to the base lines in 7 to 10 days. Mitochondrial GOT (mGOT) and glutamate dehydrogenase (GLDH; EC 1.4.1.2, -glutamate:NAD⁺ oxidoreductase) also peaked at the same time after TAE. -Fetoprotein peaked 2 h after TAE and decreased to half of the baseline on day 7. Carcinoembryonic antigen peaked at 24 h and fell at 48 h only in the patients with colon cancer. The total amount of cytosolic GOT, GPT, mGOT, and GLDH released was correlated to the volume of the necrotic mass estimated by computed tomography scans. The correlation coefficients for mGOT and GLDH were r = 0.919 and r = 0.939 (both p < 0.001), respectively. Assays of mGOT and GLDH may be useful to estimate the volume of the necrotic mass of a hepatoma or metastatic carcinoma in the liver.     

New enzymatic assay for serum urea nitrogen using urea amidolyase

We established an enzymatic assay for measurement of serum urea nitrogen using urea amidolyase (EC 3.5.1.45) from yeast species. The method is based on hydrolysis of urea by the enzyme. In this assay, we eliminated endogenous ammonium ion by use of glutamate dehydrogenase (EC 1.4.1.4). Then in the presence of urea amido-lyase, ATP, bicarbonate, magnesium, and potassium ions, ammonium ion was produced proportionally to urea concentration in serum. The concentra-tion of ammonium ion formed was determined by adding GLDH to produce NADP(+) in the presence of 2-oxoglutarate and NADPH. We then monitored the change of absorbance at 340 nm. The inhibitory effect of calcium ion on this assay was eliminated by adding glyco-letherdiamine-N, N, N', N'-tetraacetic acid to the reaction system. The with-in-assay coefficient of variations (CVs) of the present method were 1.80-3.76% (n = 10) at 2.8-19.0 mmol/L, respectively. The day-to-day CVs were 2.23-4.59%. Analytical recovery was 92-115%. The presence of ascorbic acid, bilirubin, hemoglobin, lipemic material, ammo-nium ion, or calcium ion did not affect this assay system. The correlation be-tween values obtained with the present method (y) and those by another enzy-matic method (x) was 0.997 (y = 1.02x - 0.10 mmol/L, Sy/x = 0.841, n = 100), with a mean difference of -0.18 +/- 0.86 mmol/L [(values by reference method - that of present method) +/- SD] using the Bland-Altman technique. J. Clin. Lab. Anal. 17:52-56, 2003.     

New enzymatic method with tryptophanase for determining potassium in serum.

We established a simple and rapid enzymatic method for measuring potassium ion in serum by using tryptophanase (EC 4.1.99.1) purified from Escherichia coli K12 strain (E. coli K12 IFO 3301). The presence of pyridoxal 5-phosphate promotes this enzymatic reaction, and potassium and (or) ammonium ions further accelerate it, with ammonium and potassium ions providing equivalent acceleration. We eliminated endogenous ammonium ion by using glutamate dehydrogenase (GLDH; EC 1.4.1.4), then produced ammonium ion in the presence of tryptophanase, tryptophan, and pyridoxal 5-phosphate. The concentration of formed ammonium ion, which was proportional to that of potassium ion in sample, was determined by adding GLDH to produce NADP+ in the presence of 2-oxoglutarate and NADPH; we then read the change of absorbance at 340 nm. The standard curve was linear for potassium ion concentrations up to 7.00 mmol/L. The within-assay variation (CV) was 0.89% at 5.51 mmol/L and 1.32% at 3.37 mmol/L. The day-to-day CVs were 0.99% at 6.85 mmol/L and 1.71% at 3.52 mmol/L. Analytical recoveries ranged from 98.7% to 108.9%. The correlation coefficient between values obtained with this enzymatic assay (y) and by flame photometry (x) was 0.995: y = 0.984x + 0.091 mmol/L (Sy.x = 0.105, n = 100). The presence of hemoglobin, bilirubin, or other cations little affects this system.     

Kinetic determination of serum sorbitol dehydrogenase activity with a centrifugal analyzer

We describe a mechanized method for centrifugal analyzer determination of sorbitol dehydrogenase in serum, based on conversion of D-fructose to sorbitol with simultaneous oxidation of NADH, in triethanolamine buffer at pH 7.4 and 30 degrees C. The standard curve for this assay is linear to 200 U of activity per liter of serum. The mean within-run precision (CV) of the assay is 0.8%. Results correlate well with those by a spectrophotometric method. In sera from 20 apparently healthy adult humans, sorbitol dehydrogenase activity averaged 1.7 (SD +/- 0.8; range, 1-3) U/L. The mean activity (U/L) for a group of 30 rats was 4.4 (SD, +/- 0.2; range, 3-6); for 20 dogs, 5.8 (SD, +/- 0.7; range 3-9); and for 30 mice, 26.8 (SD +/- 2.1; range, 22-34). To determine the utility of measuring this enzyme in the serum of rats for assessment of hepatotoxicity in drug-safety studies, we compared sorbitol dehydrogenase activity with that of alkaline phosphatase, aspartate aminotransferase, and alanine aminotranferase in the sera of rats treated with thioacetamide or in which the common bile duct has been ligated.     

Peroxisome localized human hepatic alanine-glyoxylate aminotransferase and its application to clinical diagnosis

The subcellular localization of alanine-glyoxylate aminotransferase (EC 2.6.1.44 L-Alanine: glyoxylate aminotransferase) of adult human liver was examined by sucrose density gradient centrifugation. The enzyme sedimented at the same density as catalase, indicating that it was localized in the peroxisomes. Alanine-glyoxylate aminotransferase activity in the liver of patients with cirrhosis was about 65% of that of normal liver or 71% of that from patients with chronic hepatitis, but its activity in the serum of patients with cirrhosis was higher than that from patients with chronic hepatitis. Patterns of activity of alanine-glyoxylate aminotransferase in liver and serum differed from those of aspartate-2-oxoglutarate aminotransferase and ornithine carbamoyltransferase that have a different intracellular location. Serum immunoreactive alanine-glyoxylate aminotransferase (Im-AGT) was measured by enzyme-linked immunoadsorbent assay (ELISA). The Im-AGT levels (mean +/- SEM) in acute (80 +/- 13 micrograms/L) and chronic (72 +/- 4 micrograms/L) hepatitis were higher than those of normal controls (44 +/- 1 micrograms/L). However, the difference between acute and chronic hepatitis was not statistically significant. The level in liver cirrhosis (54 +/- 3 micrograms/L) was lower than those of the hepatitides but higher than that of normal controls. The apparent half-life of serum Im-AGT of patients who underwent liver lobectomy by a microwave tissue coagulation method was approximately 3-4 days.     

A reference material for traceability of aspartate aminotransferase (AST) results.

Standardization of aspartate aminotransferase (AST) determination is highly desirable for inter-laboratory comparison. Serum AST mean values for 20 patients suffering from viral hepatitis showed an inter-laboratory (n = 13) variation of 9.4%. Part of this variation was due to two laboratories using procedures without pyridoxal-5'-phosphate. A traceable AST value was assigned to an enzyme calibrator (EC) through the appropriate International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) primary reference procedure. The EC was commutable for seven routine methods. Common calibration using the EC reduced the inter-laboratory coefficient of variation (CV = 5.9% ) and allowed retention of a common reference interval for a set of routine procedures. Calibration made superfluous the expression of results in multiples of the upper reference limit, which increased inter-laboratory variation (CV = 18.5%). Furthermore, for 92% of patients, calibration with the EC allowed the correction of misclassifications when taking into account the reference interval of the reference procedure. Use of this EC could be proposed to complete the AST reference system.     

An enzymatic cycling method for the measurement of myo-inositol in biological samples

INTRODUCTION: A sensitive and simple enzymatic cycling method is described for the quantitation of myo-inositol in biological samples. METHODS: The method involves the use of a sensitive and simple enzymatic cycling method is described for the quantitation of myo-inositol in biological samples. The method involves use of thio-NAD(+), NADH and thermostable myo-inositol dehydrogenase (IDH; EC. 1.1.1.18) and measurement of the increase in absorbance at 405 nm of thio-NADH at 37 degrees C. RESULTS: The calibration curve for myo-inositol was linear (r=1.00) between 10 and 400 micromol/l. Analytical recoveries of exogenous myo-inositol added to serum and urine were 100-105% and 98-103%, respectively. Within-run and between-run coefficient of variation (CV) were 0.6-2.1% and 1.1-3.0%, respectively. This method was free from interference by hemoglobin, bilirubin, ascorbate, chyle, various sugars, sugar alcohol and myo-inositol phosphates. With the use of myo-inositol as a standard solution, the serum myo-inositol concentration (mean+/-SD) was significantly greater in patients with diabetes mellitus (DM) without nephropathy (73.0+/-13.8 micromol/l, n=7) than in healthy individuals without DM (61.0+/-12.4 micromol/l, n=20). The urinary myo-inositol concentration was also significantly greater in patients with DM without nephropathy (793.3+/-870.3 micromol/l, n=7) than in healthy individuals without DM (76.0+/-63.0 micromol/l, n=13). CONCLUSIONS: This new method is simple, sensitive and enables quantitative analysis of myo-inositol.     

Production and certification of secondary enzyme reference materials (ERMs). Part 1: Preparation of the sera and some of their properties

We produced three batches of a human-serum-based enzyme reference material (ERM) enriched with human aspartate aminotransferase (EC 2.6.1.1), alanine aminotransferase (EC 2.6.1.2), creatine kinase (EC 2.7.3.2), and lactate dehydrogenase (EC 1.1.1.27). The added enzymes were not exhaustively purified; thus the final ERMs contained some enzymes as contaminants, of which only glutamate dehydrogenase activity might interfere. The stability during storage and after reconstitution was good. The commutability of the four enzymes in the three ERM batches was also good, except when German or Scandinavian methods for aminotransferases were involved. The temperature-conversion factors for the ERMs were equivalent to those for patients' sera. Reactivation after reconstitution was complete within 5 min and was independent of the temperature of the reconstitution fluid. We believe that these secondary ERMs will aid in the transfer of accuracy between well-defined reference methods and daily working methods so that clinical enzymology results will become more comparable from laboratory to laboratory.     

Kinetic assay of gamma-glutamyltransferase with use of bilirubin oxidase as a coupled enzyme

We studied the kinetic measurement of gamma-glutamyltransferase (EC.2.3.2.2), coupling the reaction with that catalyzed by bilirubin oxidase (EC 1.3.3.5), which oxidizes and combines a phenylenediamine derivative with an aniline derivative to produce a green pigment. We measured the formation of the pigment kinetically (at lambda max745 nm, epsilon = 75 000 L mol-1 cm-1), with L-gamma-glutamyl-N-hydroxyethylaminoanilide as substrate and N-ethyl-N-hydroxy-3-sulfopropyl)-m-toluidine as a the coupling derivative. The within-run CV for measuring this reaction in samples of normal sera was 2.4%. A calibration plot of the change in absorbance per minute vs enzyme activity concentration showed good proportionality in the range of 0-1300 U/L. The results of this assay correlated well (r = 0.995) with those of the Boehringer method, in which L-gamma-glutamyl-3-carboxy-4-nitroanilide is the substrate. This new, highly sensitive procedure may be adapted to other assays involving phenylenediamine derivates as synthetic substrates.     

Activities of aspartate and alanine aminotransferases and alkaline phosphatase in sera of healthy subjects.

The activities of serum aspartate aminotransferase (EC 2.6.1.1, L-aspartate: 2-oxoglutartate aminotransferase, ASAT) and alanine aminotransferase (EC 2.6.1.2, L-alanine: 2-oxoglutarate aminotransferase, ALAT) were determined in the sera of 1484 apparently healthy subjects using kinetic methods according to the Scandinavian recommendation (33). In the adult sera the mean activity of ASAT was 21.4     

A kinetic procedure for the estimation of arginine in serum using arginine kinase

A method for estimation of arginine in 50 microliters serum was developed using commercially available arginine kinase (EC 2.7.3.3). The assay is based on the transformation of arginine and ATP into phospho-arginine and ADP by the enzyme. ADP is measured by two coupling reactions involving pyruvate kinase (EC 2.7.1.40) and lactate dehydrogenase (EC 1.1.1.27) with measurement of NADH consumption at 340 nm. The method involves preincubation of serum in the reaction medium without arginine kinase to eliminate side reactions and a kinetic rate protocol with measurements of absorbance at 60 s and 180 s. Reaction temperature is 30 degrees C. The reaction is linear up to at least 3 mmol/l of arginine. Within-batch CV is less than 3% for arginine levels above 0.75 mmol/l and the between-batch CV is 6.5% or less. The method correlates well with an automatic amino acid analyzer procedure (r = 0.983). The reference range derived from sera of 40 blood donors has been determined to be 0.06-0.20 mmol/l.     

Interference by endogenous glycerol in an enzymatic assay of phosphatidylglycerol in amniotic fluid

We investigated the possibility of interference by endogenous glycerol with the enzymatic measurement of phosphatidylglycerol in amniotic fluid. Phosphatidylglycerol is an important indicator of fetal lung maturity. The concentrations of glycerol and phosphatidylglycerol in amniotic fluid were measured by using a coupled enzymatic assay with and without phospholipase D (EC 3.1.4.4). The precision of the assay was acceptable (within-run CV = 1.2%, between-run CV = 4.8%). Endogenous glycerol content was demonstrated to be approximately 10-20 times that of phosphatidylglycerol. This high proportion of endogenous glycerol in amniotic fluid would preclude the accurate enzymatic determination of amniotic fluid phosphatidylglycerol unless the glycerol is first removed. Nor can the actual phosphatidylglycerol concentration be determined by subtracting the endogenous glycerol concentration from the total glycerol, which includes that glycerol derived from phosphatidylglycerol. With a usual range of 9 +/- 7 mumol/L, the error for a given phosphatidylglycerol measurement of +/- 6.6 mumol/L (+/- 2 SD) clearly is too high for this assay to be clinically useful. There was no correlation between concentration of endogenous glycerol or apparent phosphatidylglycerol in amniotic fluid and the lecithin/sphingomyelin ratio of the sample.     

A sensitive, direct colorimetric assay of serum iron using the chromogen, nitro-PAPS

A direct colorimetric method is presented for the determination of serum iron in 0.1-ml sized samples, using a new, water-soluble, reagent, 2-(5-Nitro-2-pyridylazo)-5-(N-propyl-N-sulfopropylamino)phenol Na salt (epsilon 585 nm = 9.4 X 10(4) l/mol per cm). Interference of copper and zinc in sera can be eliminated entirely by forming copper- and zinc-thioglycollate complexes immediately upon the dissociation of the protein-bound iron, copper and zinc by thioglycollate and sodium dodecyl sulfate. The serum blank was minimized by the use of sodium dodecyl sulfate as a protein denaturant. Within-run and between-run precision (CV) were in the range of 0.7-2.9% and 1.1-3.6%, respectively, depending on the serum iron content. A good correlation (r = 0.995) was obtained between this method and the reference method proposed by the International Committee for Standardization in Hematology.     

Improvement of a direct spectrophotometric assay for routine determination of superoxide dismutase activity.

The growing interest in measuring superoxide dismutase (EC 1.15.1.1) in many diseases calls for useful routine assays. For this purpose, the direct spectrophotometric method of Marklund (J Biol Chem 1976;251:7504-7) was improved to offer an alternative to the imprecise, indirect assays currently used. The decay of O2.- (from KO2) at pH 9.5 was monitored as the decrease in delta A (delta A = A250nm-A360nm). Superoxide dismutase was determined from the pseudo-first-order rate constant of O2.- dismutation. The precision of the assay was improved by increasing the concentration of O2.- and expanding the interval for measurements of O2.- concentrations to 4-16 mumol/L. Other assay characteristics, including temperature, were also optimized. In hemolysate the assay had a within-day CV of 5.5-13% and a between-day CV of 4%. Mn-superoxide dismutase and some superoxide dismutase mimics are inhibited at alkaline pH. Therefore, the method is primarily recommended for Cu,Zn-superoxide dismutase.     

A spectrophotometric micromethod for determining erythrocyte protoporphyrin-IX in whole blood or erythrocytes.

An increased concentration of erythrocyte protoporphyrin-IX in whole blood or erythrocytes is a valuable diagnostic indicator for acquired porphyrias, e.g., iron deficiency anemia and lead poisoning, and for inherited porphyrias. We developed a spectrophotometric micromethod for determining erythrocyte protoporphyrin-IX. In this method, exhaustive release of erythrocyte porphyrins is achieved using hydroquinone and formic acid. The clean-up procedure for 50 microL of whole blood or erythrocytes covers three steps of liquid/liquid solvent partition: two partitions using diethyl and diisopropyl ether and HCl 2.5 mol/L, and one buffered step using ammonium formate. Determinations of erythrocyte protoporphyrin-IX are possible by: (a) absorption using three wavelengths, Rimington's constant and a millimolar absorptivity coefficient m epsilon(408.8)=294.3 L x mmol(-1) x cm(-1) according to With; and (b) 2nd derivative, which is linked to m epsilon(408.8). Determination of erythrocyte protoporphyrin-IX using a 2nd derivative algorithm showed better spectral resolution and higher sensitivity at a five-fold lower detection limit compared to absorption. Within-run precision of medium and high levels was found for absorption and for 2nd derivative with a coefficient of variation (CV) of 1.4-1.9% (n=10). Total precision evaluated was CV=2.5-8.3% (n=20). Levels of reference intervals could only be measured using the 2nd derivative (CV 2.9%). Linearity was proved to E=1.0. Recoveries of protoporphyrin-IX ranged from 95.3% to 103.0%. Method comparison was carried out using a fluorimetric reference method (Piomelli). Reference intervals for gender groups are discussed.     

Bioluminescence assay of creatine kinase and its isoenzymes in serum and cerebrospinal fluid.

We examined the sensitivity of bioluminescence for the determination of very low concentrations of creatine kinase brain-type subunit (CK-BB) in serum and in cerebrospinal fluid. To optimize the sensitivity of CK-isoenzyme assays and eliminate possible sources of error, we separated the isoenzyme fractions by using inhibiting anti-MM and precipitating anti-MM and anti-BB antibodies. The results with the bioluminescence assay correlated with spectrophotometric values such that r = 0.97 for the total CK activity and r = 0.98 for the CK-B activity. The reproducibility of the present method was comparable with the spectrophotometric method and was even better at low enzyme activities. The within-series precision for assay of total CK activity at 2 U/L corresponded to a CV of 9%; at 13 U/L the CV was 5.8%. All the assays were carried out at 25 degrees C. Even at this low temperature, CK activities as low as 0.2 U/L could be determined. In eight patients without any evidence of cerebral cell damage, total CK activity in cerebrospinal fluid was x = 1.05 +/- 0.6 U/L, and CK-BB activity was x = 0.7 +/- 0.4 U/L. In sera of these patients CK-BB activity was x = 0.6 +/- 0.5 U/L. Differences in CK and CK-BB activities in four patients with transient or progressive brain-cell damage are discussed.     

New system of continuous monitoring of enzyme activities and determination of some substrates

The proposed system of continuous monitoring of enzyme activities is based primarily on the electrochemical behaviour of thiol compounds, and the experimental equipment is extremely simple. The determination of cholinesterase (EC 3.1.1.8) activity is described. The normal values obtained for men (73.9, s +/- 10.3 microkat/l) and for women (71.1, s +/- 10.2 microkat/l), lie within the usual range of analogous photometric methods. Systems for determination of the activities of alkaline phosphatase (EC 3.1.3.1) and adenosylhomocysteinase (EC 3.3.1.1) are described. The activity of aspartate aminotransferase (EC 2.6.1.1) is determined by a combination of enzyme reactions, in which CoA is released from acetyl-CoA. Analogous procedures are discussed for determinations of alanine aminotransferase (EC 2.6.1.2), lactate dehydrogenase (EC 1.1.1.27), lipase (EC 3.1.1.2), and phospholipase A2 (EC 3.1.1.4) activities, and for determination of substrates, e.g., acetate and carnitine. Possible determinations of an additional 199 enzyme activities and of some of substrates are suggested. By determining electrochemically active groups other than thiols this method becomes almost universally applicable.     

Evaluation of a radioimmunoassay for estradiol in unextracted serum

We describe the performance of a commercial (Steranti/EIR) RIA reagent kit for measuring 17 beta-estradiol directly in serum. Day-to-day precision data for control sera were as follows: mean = 102.8 ng/L, CV = 6.8%, n = 20; mean = 231.1 ng/L, CV = 5.3%, n = 21; mean = 747.7 ng/L, CV = 9.4%, n = 21. Analytical recovery of added estradiol from seven different serum pools from men, to which three different concentrations of estradiol had been added, was (mean +/- SD): 98.6 +/- 7.0% at 107.5 ng/L added; 98.8 +/- 4.7% at 322.5 ng/L added; 108.2 +/- 4.8% at 645 ng/L added. Overall recovery of estradiol in these experiments (mean +/- SD for 21 determinations) averaged 101.9 +/- 7.0%. Assay of 32 serum specimens from women by both the direct (y) and an extraction method (x) gave the following linear regression statistics: y = 1.12x - 1.3, r = 0.998, Sy/x = 30.2 ng/L, mean y = 438.2 ng/L, mean x = 391.4 ng/L. Hemoglobin, bilirubin, and moderate lipemia do not interfere. Sensitivity of the direct assay was 2.6 ng/L. Compared with the extraction assay, the direct estradiol assay has advantages of speed and simplicity.