coated tubes(Becton Dickinson and Coat-A-Count), magnet- ic separation (Amerlex Magnetic and Coming Magic), or centrifugation of antibodies linked to solid ...
CLIN. CHEM. 31/11, 1888-1892 (1985)
Accuracyand Precisionof Five AnalogRadioimmunoassays for Free ThyroxinCompared S. M. Gow,1 H. A. Kellett,2 A. D. Toft,2 and G. J. Beckett1 We compared the precision of free thyroxin (FT4) measurements by kits involving analog AlA and the use of antibodycoated tubes (Becton Dickinson and Coat-A-Count), magnetic separation (Amerlex Magnetic and Coming Magic), or centrifugation of antibodies linked to solid beads (Amerlex). Results of kits with magnetic separation were the most reproducible. Amerlex, Amerlex Magnetic, and Becton Dickinson kits gave values comparing best with those obtained by direct equilibrium dialysis. Coat-A-Count and Coming Magic results differed significantly from dialysis values, both for
compared these assays for (a) precision, (b) accuracy as compared with a reference dialysis method, and (c) diagnostic efficiency in patients with suspected thyroid disease, in pregnant women, and in patients with nonthyroidal illness.
Materials and Methods FT4
by Equilibrium Dialysis
Samples were diluted 20-fold with pH 7.4 buffer contain-
patients’ samples and kit standards. The kits had equal diagnostic efficiency in patients with suspected thyroiddis-
ing, per liter, 10 mmol of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (ssxpzs;Sigma Chemical Co.), 106 mmol of NaC1, and 1 mmol of NaN3, then dialyzed against an
ease. On measurement of FT4 some patients were reclassified from “subclinical thyroid disease” to “overt disease.” Most patients with triiodothyronine thyrotoxicosis had increased FT4. Several kit values were low for pregnant women and patients with nonthyroidal illness but the Amerlex and Amerlex Magnetic assays had fewer low results. The Amerlex Magnetic FT4 assay gave the best precision, agreement with the reference method, and diagnostic efficiency.
equal volume of this buffer at 37#{176}C in a #{176}Dianorm” equilibrium dialyzer (Diachema AG, Zurich, Switzerland) for at least 18 h. The T4 in the dialysates was then measured in duplicate by RIA [FF4(D)] with a pre-precipitated doubleantibody system (8). Between-run precision was assessedby dialysis and RIA of three serum pools in 15 assays. The mean precision (CV) was 11% for Fr4 values in the range 7 to 27 pmoIIL.
AddItional Keyphrasea: thyroid status variation, source of pregnancy “kit”methods diagnostic efficiency
FT4 by Analog AlA
‘
‘
Estimation of free thyroxin (FF) in serum is generally considered to provide a more nearly accurate diagnostic test of thyroid dysfunction than is measurement of total thyroxin (T4) in serum or determination of the free thyroxin index (FF1), particularly in conditionswhere there are changes in the concentrations of thyroid hormone-binding proteins such as in pregnancy (1-3). It is also currently held that only the free fraction of thyroid hormones can gain access to cells and thus exert metabolic effects (4,5). The development by Amersham International of the one-step RIA involving a radiolabeled analog of thyroxin (6) has resulted in FF4 assays that are more precise and robust than earlier commercial methods involving microencapsulated antibodies and two-step kinetic rate analysis (7). These advantages have led many companies to adopt this methodology, and several Fr4 RIA kits now marketed are based on the useof different T4 analogs and separation systems. Many studies evaluating individual FF4 kits have compared them with the Fl’! and other thyroid-function tests rather than with a reference method for FF4. Equilibrium dialysis of serum followed by RIA ofT4 in the dialysate is the reference method for FF but is unsuitable for routine use (5). Here we report our assessment of five different FF4 analog assays having various separation systems. We have
We studied Fr4 kits from the following manufacturers: Amerlex and Amerlex Magnetic (Amerlex-M) (Amersham International, Amersham, Bucks., U.K.); Becton Dickinson (Becton Dickinson UK Ltd., Cowley, U.K.); Coat-A-Count (Diagnostic Products (UK) Ltd., Wallingford, U.K.); and Corning Magic (Corning Medical and Scientific, Haistead, U.K.). The Amerlex-M kit, which contains the same analog and standards as the Anierlex kit, became available after the start of this study. Table 1 summarizes the assay protocols. All assays were performed by one operator, using an automatic diluter and dispensing sample with tracer in one step within 10 mm to minimize effects of drift, particularly in the coated-tube assays. Subjects Endocrine clinic patients. Samples were collected from 200 consecutive patients referred to an endocrine clinic.
These patients were categorized on the basis of clinical examination by one consultant, and results of measurements of serum T4, total triiodothyronine (T3), basal thyrotropin (TSH), and the TSH response 20 mm after intrave-
Table 1. Comparison of FT4 Analog RIAs Am.rI.x FT4
Samplevol, University Departments of’ Clinical Chemistry and2 Medicine, The Royal Infirmary, Edinburgh EH3 9YW, Scotland, U.K. 3Nonstandard abbreviations: ?1’4, free thyroxin; FF1, free thyroxin index; T4, total thyroxin; FF4(D),FF as measured by RIA after dialysisto equilibrium; TSH, thyrotropin (thyroid-stimulating hormone); TRF, thyroliberin (thyrotropin-releasing factor); and T3, triiodothyronine. ReceivedMay 8, 1985; acceptedAugust 1, 1985. 1888
CLINICAL CHEMISTRY, Vol. 31, No. 11, 1985
AIeerI.X
Msgn.tlc
100
100
BectonDickinson 50
Coat-ACount 50
Coming M.gic 50
&L
Pipetling 2 2 1 steps Incubationat 1 1 1.5 37#{176}C, h Wash step No No Yes Sepn. Centrifn. Magnetic Coated method
tube
1
2
1
1
No
No
Coated tube
Magnetic
nous administration of 200 pg of TRF (TRF stimulation test). The results for 63 patients classified clinically and biochemically as euthyroid were normally distributed. Weused them to derive reference intervals (mean ± 2SD) for each FF4 method. The 110 hyperthyroid patients had increased concentrations of total thyroid hormones and a TSH responseof 0
if.
12 10
>6 0
3
6
125
25
50
20
100
FREE TL (pmol/LJ
Results Precision: Figure 1 illustrates the mean within-assay precision profiles for 15 assays by each method. These profiles were derived from analysis of the duplicate measurements for samples (9). The relatively poor precision of the dialysis method represents the combination of imprecision originating from the dialysis and RIA steps. The Amerlex and Amerlex-M assays had the lowest profiles compared with the coated-tube assays. The Corning Magic kit was less precise at low values. For practical purposes the precision attained with all of the kits was adequate between 3 and 100 pmoIJL. We assessed the between-assay precision for the analog FF4 methods, using two control sera (luAmAc ii and m, Becton Dickinson) and a low-concentrationserum pool (Table 2). In addition, a patient’s sample from the previous run was re-analyzed in the next assay (repeat-analysis control) and, for values within the range 5 to 50 pmol/L, the precision (CV) calculated from the two results on each repeat-analysis control from 10 assays was: Amerlex 9%, Amerlex-M 5%, Becton Dickinson 9%, Coat-A-Count 7%, and Corning Magic 6%. Each test was performed in duplicate, sowe compared the average number of tests per run where there was less good agreement between replicates (duplicate error). A duplicate error was indicated if the variance ratio of the replicate counts to the mean counts exceeded the arbitrary value of 20 (9). For each kit the mean percentage of duplicate errors per run was: Amerlex 27%, Amerlex-M 14%, Becton Dickinson 18%, Coat-A-Count 24%, Corning Magic 13%. This represents the number of tests where the replicate values required further scrutiny to determine if re-analysis was necessary. Drift analysis: One sample in each of 10 assays was analyzed at the beginning and the end of batches of 50
Fig. 1. Mean within-run CVs for 15 assaysof FT4by each method Reference intenelswere: 11-23 CoatA Count);8-17 (S, BectonDickinson, andA, eilllbdum dialysIs); 10-22 (#{149}, AmerlexMagnetIc);10-21(0,Amedex); 17-30(x,Coming)pmot/L Ba,z indIcate1 SD
(,
Table 2. Between-Assay Precision of the FT4 Kits Amerlx
Low-concn.pool n 13 Mean, pmol/L 7
Amedex Becton- Cost-A- Coming Magnetic Dickinson Count Magic 16 7
15 7
10 8
19 13
SD, pmol/L
0.3
0.3
0.6
0.6
0.8
CV,%
5
5
9
7
6
RIATRAC
H
n
17
19
15
10
24
Moan, pmoVL
12
13
9
11
19
SD, pmol/L
0.6
0.6
0.8
0.8
0.9
CV,%
5
5
9
7
5
15 35
16
16
10
24
31
21
28
38
RIATRAC
III
n Mean, pmovL SD, pmol/L
CV,% and iii
RIATRAC ii
2.2 2.1 7 10 4 are commercialcontrolsera. 1.4
2.0
1.9
7
5
samples, each run in duplicate. There was no significant drift in the Amerlex, Amerlex-M, and Corning Magic assays, but both coated-tube assays showed a positive drift of 7 to 8% (p
