Characterization of an Automated Radioimmunoassay for T4, T3, T3U, and FTI

Transcription

1 ANNALS OF CLINICAL AND LABORATORY SCIENCE, Vol. 1, No. 1 Copyright 198, Institute for Clinical Science, Inc. Characterization of an Automated Radioimmunoassay for T4, T3, T3U, and FTI ROLAND VALDES, Jr., Ph.D.* and JOHN T. USETED, B.S., M.T. The Jewish Hospital of St. Louis Washington University School of Medicine, Department of Pathology and Laboratory Medicine, St. Louis, MO ABSTRACT The performance characteristics of assays is reported for thyroxine (T4), triiodothyronine (T3), and T3-uptake (T3U) using the GAMMAFLO Automated Assay System. A comparison of calculated free thyroxine index (FTI) values is also presented. This automated radioimmunoassay (RIA) system utilizes a combination of continuous-flow methodology and chromatographic separation techniques. The T4 assay studied had a standard curve range of 1.5 to 4.0 fig per dl. The intra- and inter-assay precisions were 4.3 and 5.3 percent CV, respectively, for a T4 concentration of 10.0 fxg per dl. The T3 assay had a standard curve range of 50 to 1000 ng per dl, the corresponding precisions were 7.3 and 7.1 percent CV, respectively, for a concentration of 13 ng per dl. The automated serum T4 and T3 results correlated (r = and 0.864) with a manual radioimmunoassay procedure. Intra-assay and inter-assay precisions for a mid-range normal 30.1 percent T3U value were 6. percent and 4.9 percent CV, respectively. Reference range comparison of FTI by both automated and manual results correlated for 47 out of 51 (95 percent) patients compared. It is concluded that this automated system appears to offer a viable alternative to T4, T 3, and T 3U manual RIA techniques in terms of operational simplicity, analytical performance, and sample through-put flexibility. Introduction the counting of radioactivity. Methods for A major problem in the automation of the automation of RIA have been describ ed using both d iscrete and RIA procedures has been the separation of the bound from the free radiolabelled continuous-flow systems.4,6,7 The Gamfractions and the coupling of that step to maflow concept1combined the features of continuous-flow analysis with those of * Author to whom correspondence should be discrete sample systems and chromatographic techniques for separation of the addressed /8/ $01.50 Institute for Clinical Science, Inc.

2 A U T O M A T E D R A D IO IM M U N O A SSA Y F O R T 4, T 3, T 3U, A N D F T I 4 3 bound from free radiolabelled ligand. In th a t system, in te rru p tio n of flow (continuous-flow/stopped-flow) provided the capability for chromatographic separation and stationary counting of the bound radioactive fraction. The study reported here was undertaken to evaluate an automated assay for serum T4 (L-thyroxine), T3 (triiodothyronine) and T3U (triiodothyronine uptake) using the GAMMAFLO Automated Assay System. The results of calculated free thyroxine index (FTI) were compared by both automated and manual methods. Materials and Methods A p p a r a t u s The GAM M AFLO A utom ated Assay System* was used. The system consisted of the components described previously.9 Sampling and mixing of reagents were accomplished by a movable pipet arm with three aspiration probes to draw the sample, antibody, and radioligand simultaneously from their respective cups. Since reactants were introduced simultaneously with each sample, there was no reagent wastage between samples. The reaction mixture was incubated in a delay coil submerged in a thermostated water bath at a temperature of 37 ± 0.5. After an incubation time of six minutes and 0 seconds, the radioactivity in the mixture itself triggered a preset timed valve sequence. This sequence controlled the sample flow through the separating and counting portions of the system. Separation of free from bound radioligand was accomplished by an ion exchange resin separating column ( ml bed volume) supplied with each kit. Counting of the radioactive fraction (antibody bound fraction for the T4 and T3 assays and fraction bound to endogenous protein for the T3U assay) was performed by an interrupted-flow technique which allowed for stationary counting of this bound fraction and its subsequent rapid elimination from the detection system by vacuum aspiration. The instrum ent was com puter-controlled, was fully automated, and had available several modes for data reduction. R e a g e n t s Autom ated Assays. All reagents for the automated T4, T3 and T3U assays were used as supplied by the manufacturer.* The RIA kits for the T4 and T3U assays were commercially available.* The T 3 RIA material was obtained from the same manufacturer while still undergoing field evaluation. A variety of different kit lot numbers were used throughout this study. For the T3-uptake assays, in place of the antiserum, a T3U diluent (albumin based matrix) was supplied in 10 ml quantities. Lyophilized controls were obtained from the manufacturer.! The controls were reconstituted in 5 ml deionized H0. The T4 and T3 working buffer was prepared by dilution of 100 ml of a concentrated sodium barbital-tris buffer* to a final volume of 100 ml with deionized H0 and stored at room temperature. The concentrated buffer had a ph of 8.6 and contained Rrij surfactant. The column wash solution was also supplied by the manufacturer and stored at room temperature. The T3U buffer was the same as the previous buffer except that 100 ml of concentrated buffer was diluted to a final volume of 500 ml with deionized H 0. The radioactive tracer for these three assays was Iodine- 15. Unless otherwise stated, all reagents and standards for the automated assays were supplied by E. R. Squibb and Sons, Inc. in kit form, stored at 4 and equilibrated to room temperature prior to use. * E. R. Squibb and Sons, Inc., Princeton, NJ f BIO-RAD Laboratories, Anaheim, CA 9806, (Lypho-check, ECS, radioassay control serum, human).

3 4 4 V A L D E S A N D U S E T E D TABLE I Gammaflow Thyroid Function Test Assay Conditions* Thyroxine Triiodothyronine$ Tj Uptake Incubation temperature Incubation time Separation column material Buffers and reagents Sample and reagent draw time (volume) Interval wash draw time (volume) Concentration of standards Standard replicatesf Sample replicates Standard curve fitting routine Standard curve printout Counting time Cup loading sequence Determinations per hour 37 ± 0.5 C 6 min 0 s Cellulose ion exchange resin (See materials and methods) 3.4 s (1 ul) 6.6 s (918 UL) 0, 1.5, 3.0, 6.0, 1.0, 4.0 (ug/dl) 3rd order polynomial Bound counts vs. Ln (dose) 7 s ] - 6 standards 7-N samples and controls ± 0.5 C 6 min 0 s Cellulose ion exchange resin (See materials and methods) 4.9 s (17 \il) 85.1 s (1,50 yl) 0, 50, 100, , 1000 (ng/dl) 3rd order polynominal Bound counts vs. Ln (dose) 33 s 1-6 standards 7-N samples and controls ± 0.5 C 6 min 0 s Cellulose ion exchange resin (See materials and methods) 4.9 s (17 ul) 67.1 s (984 UL) Normal, 30 percent 5 Percent calculation None 3 s 1 standard 3-N samples and controls 50 *Conditions are pre-programmed (except for number of sample replicates) as supplied by the manufacturer and may be changed as necessary. $The sample and antiserum pump tubes are interchanged for this assay only. freplicate number is the number of determinations made from one cup, each determination and their arithmetic means are printed. This is specified by the operator with a value of up to 5 permissible. Manual Assays. Manual assays for T4, and T3 and T3U were performed. $ The T4 and T3 kits used 15I-tagged ligands and a second antibody plus polyethylene glycol (PEG) for separation of bound from free ligands. Charcoal bound counts of 15I-T3 were measured in the T3U assay. P r o c e d u r e s The automated assays of T4, Tg, and T3U were performed using the conditions detailed in table I. A fresh column (approximately ml bed volume) was used as supplied for each kit. Analysis was initiated by keyboard entry of the assay type (T4, T3, or T3U) and the number of sample replicate determ inations desired. T he num ber of replicate determinations per sample cup varied depending on the nature of the experiment. All Gammaflo kits } Diagnostic Products (DP) Assay Kits (Diagnostic Products Corporation, Los Angeles, CA 90064). were used as provided by the manufacturer with only the standards having to be loaded into sample cups, i.e., the antibody and tracer materials were aspirated directly from their package vials. The T3-uptake measurements were performed by aspirating the T3U diluent (in place of the usual antibody), the radioligand 15I-T3, and sample, simultaneously as in the usual automated procedure. Passage of this reaction mixture through the column resin separated the free radioactive T3 from that bound to the endogenous protein in the specimen by allowing the bound fraction to pass into the counting Teflon-well while the resin binds the free residual 15I-T3. D a t a R e d u c t i o n The curve fitting routines available on the autom ated system w ere logit-log, w eighted logit-log, adjusted end-point

4 A U T O M A T E D R A D IO IM M U N O A SSA Y F O R T 4, T 3, T 3U, A N D F T I 4 5 logit-log, iterative log fit, third order polynomial, spline, and a relative percent calculation. In the present study, the third order polynomial regression was used for all the T4 and T3 determinations. This routine fitted the number of bound counts against the square root of the analyte concentration (dose). A standard curve was printed followed by a printout of the sample and control values in B (bound counts), B/B ratio (B0 represents total bound count for a zero concentration standard) and pre-selected concentration units. An additional feature was the ability to use the curve fitting routines in a manual data entry and computation mode. The T3U m easurem ents were calculated using a relative percent calculation program supplied by the manufacturer which used a normal standard as a reference value: marized in table II. In order to simulate routine laboratory conditions, cups containing serum samples from patients were assayed b etw een sets of control cup positions for the intra-assay studies. Inter-assay precision was determ ined by analyzing the three BIO-RAD control samples over the number of separate automated assay runs indicated, with duplicate determinations each run; these results are also tabulated in table II. New bottles of the controls were prepared and used each assay in order to include vialto-vial variability. Carry-over. Sample carry-over3,9 calculated for the T3U assay was 0. percent for the high and low T3U values of 38 percent and 19 percent assayed consecutively and based on single determinations. Corresponding T4 RIA carry-over was 0.8 per- (Avg. Normal Count) ( ) - Sample Count Percent Uptake = ^ (y) 100 (Avg. Normal Count) ) - Avg. Normal Count where y = the defined percent uptake value of the normal standard. Data reduction for the manual assays were performed by our laboratory mainframe* utilizing a weighted logit-log calculation for the T4 and T3 assays and a percentage calculation for the T3U assay defined as: cent for the two thyroxine values of and 6.18 fig per dl. The T3 RIA carry-over was measured as 1.8 percent for the two consecutive values of and ng per dl. These carry-over results are indistinguishable from zero based on the assay precisions. Percent binding = (avera e s.^!pp^e counts) x (percent uptake assigned to calibrator), (average calibrator counts) Results P r e c i s i o n Intra-assay precision was determined using three BIO-RAD control samples by making a number of consecutive determinations at each concentration. The results for the T4, T3, and T3U assays are sum- * Technicon LDM system. Drift. Assay drift was assessed by trend analysis as previously described9,10 by measuring three control samples for 150 consecutive determinations for each of these three assays. In all cases, the slopes obtained (y = measured value, x = determination number) were indistinguishable from zero within the precision described. An example of a drift and carry-over assay experiment is shown in figure 1.

5 4 6 V A L D E S A N D U S E T E D TABLE II Precision of Automated Assays No. of Determinations N Mean ± S.D. Our Measured Percent cv Manufacturer's* Ci ted Percent CV Thyroxine ( uq/dl) Intra-assay Inter-assay Triiodothyronine (nq/dl) Intra-assay NA NA NA Inter-assay NA NA NA Uptake ("percent T?J) Intra-assay NA Inter-assay NA Manufacturer's cited CV's for comparable concentration range. NA = Not available CARRYOVER A N D DRIFT DETERM INATIO N NUMBER F i g u r e 1. A typical carry-over and drift experiment. The T4 assay is depicted here monitoring thyroxine concentrations of.1 (L), 6. (M), and 17.7 (H) Jig per dl. Each cluster of 10 points represents ten consecutive determinations from a sample cup. The cups are assayed consecutively in the order L, M, H, L, M, H, etc. starting with determination number 15. Carry-over is assessed between the H and L cups, see text for values. A c c u r a c y Accuracy for the T4 and T3 assays was assessed by recovery, cross reactivity, dilution, and patient sam ple correlation studies. Recovery of the thyroxine assay was determ ined by the recovery of w eighed amounts of L-thyroxinef which had been pre-dissolved in ethanol and ammonium hydroxide (4:1 v/v). This solution was added quantitatively to an albumin matrix solution (T3U diluent) and this later solution subsequently added to a patient serum pool. D uplicate analyses were made on samples from this pool with added thyroxine concentrations of 5.0, 10.0, and 15.0 fig per dl; three samples were prepared at each concentration (triplicate spikings). The recoveries were 99, 108, and 116 percent for these three concentrations, respectively, with a mean of 107 percent. The recovery for the T3 RIA was determ ined with a similar protocol using 3,3 ',5 -triio d o -L -th y ro n in ef at added concentrations of 100, 00, and 300 ng per dl T3 and measured recoveries of 105, 104 and 101 percent, respectively, with a mean T3 recovery of 103 percent. Cross reactivity studies of the T4 assay for T3 were performed by preparing a set of T3 (3,3',5-triiodo-L-thyronine) samples pre-dissolved as described before for the T4 recoveries. Twelve T3 concentrations were prepared ranging from 6.5 to 1000 fig per dl. The 50 percent 15I-T4 ligand displacem ent corresponded to concentrations of 10.6 fig per dl for T4 (standard curve) and 75 fig per dl for T3 giving a 14.1 percent cross-reactivity of T3 for the T4 assay based on equivalent mass concentrations.5 The autom ated T3 assay was evaluated for cross-reactivity with T4 and reverse T3 (rt3). The T4 solutions were prepared as previously described at concentrations ranging from 7.5 to 150 fig per dl. The 50 percent B/B0 displacement f L = thyroxine, sodium salt, pentahydrate; Sigma Chemical Co., St. Louis, MO.

6 A U T O M A T E D R A D IO IM M U N O A SS A Y F O R T 4, T 3, T 3U, A N D F T I 4 7 corresponded to concentrations of 548. ng per dl T3 and 667 xg per dl for T4 yielding a 0.08 percent cross-reactivity of T4 for the T3 assay. An identical study using rt3 (3,3',5'-triiodo-L-thyronine)f gave resu lts of 0.04 p e rc e n t crossreactivity of rt3 with the T3 RIA. These results are summarized in table III with the effective physiologic calculated cross reactivity indicated for T3 in the T4 assay based on an average 50-fold difference in circulating physiologic concentrations of these two analytes. D ilution studies were performed on both T4 and T3 assays over a range of concentration covering the dynamic range of their respective standard curves. In T4 dilution experiments, a patient serum with an initial T4 concentration of 17.4 fxg per dl was diluted with zero standard to a low concentration of 1.1 /xg per dl. The linearity observed was y (expected) = x (observed) 0.7 /xg per dl, (n = 10) Sey = 0.1 / g per dl, r = Two T3 dilution experiments were performed by diluting with zero standard and a low patient serum pool, respectively. Both dilution curves were linear and essentially identical with average parameters of y = 1.00 x ng per dl, (n = 9) Sey = 1. ng per dl, r = For both assays, the dilution was indeed linear with intercepts approaching zero demonstrating parellelism. The linear results of the dilution studies for both T4 and T3 assays are shown in figure. P atient C orrelation. T he thyroxine concentrations in 74 patient samples were measured using the manual RIA procedure* and the results compared with those ob tained by th e GAM M A FLO system. These results with corresponding parameters are shown in figure 3. The patient correlation of the GAMMAFLO T3 assay with the DP manual assay is shown in figure 4 for 46 patients. f L-thyroxine, sodium salt, pentahydrate; Sigma Chemical Co., St. Louis, MO. * Diagnostic Products. TABLE I I I Automated Thyroid Function Tests: Cross Reactivity Assay Cross Reactant (50 Percent Dispiacement) Observed* Percent Cross Reactivity t4 T4 100 T3 14.1$ t3 T3 100 t4 0.08f rt *No 50 percent displacement cross reactivity cited in manufacturer's (Gammaflo) package insert for comparison. $The 14.1 percent cross reactivity of T 3 for the T4 assay is an effective 0.3 percent cross reactivity when normalized to relative endogenous serum concentrations. This value is lower than that reported for the DP manual kit (0. 8 percent). T4 cross reactivity with the T 3 assay (0.08 percent) was lower for the Gammaflo than that cited for the manual assay (0. 6 percent). The patient correlation study for the T3-uptake test is shown in figure 5. The regression analysis gave a correlation coefficient of0.793 with means of x = 5.5 and y = 8.0 percent T3U. However, because of the clustering of values near the normal range, it was felt that a correlation analysis by the categories hypo-, normal, and hyper- would yield a more meaning- DILUTION F ig u r e. D ilu tio n stu d y fo r T 4 a n d T 3 a u to m a te d a ssa y s. S e ru m sá m p le s d ilu te d w ith z e ro sta n d a rd s h o w n h e re. D ilu tio n w ith p o o le d s e ru m gave co m p arab le results.

7 48 V A LD ES A N D U S E T E D Calculation of the free thyroxine index (FTI = T4 x T3U) using the T4 and T3U values obtained with the GAMMAFLO and manual kit assays for the same 51 patients gave the correlation depicted in figure 6. W hen considering both parameters, T4 and T3U as in the FTI calculation, four patient out of fifty-one (8 percent) were discrepant in biochemically classifying the patients. These four patients were classified within the normal range by the manual kit but hypothyroid by the GAMMAFLO assay. F i g u r e 3. Patient correlation for T 4 assays. Mean values were x = 8.43 and y = 9.05 ng per dl. ful interpretation. Of the 51 patient samples, 8 percent were coincident for all three categories; nine samples (8 percent) were discrepant, of which five (10 percent) were classified as normal T3U by the GAMMAFLO assay and hyper by the manual determination. The remaining four discrepant samples (6 percent) were classified as hypo by the automated method and normal by the manual determination. F igure 4. Patient correlation for T3 assays. Mean values were x = 11 and y = 114 ng per dl. Discussion Thyroxine, triiodothyronine and T3- uptake assays performed on a fully autom ated radioim m unoassay system have been evaluated and compared to a manual RIA. In addition, the calculated free thyroxine index for both the automated and the manual assay were compared. The performance characteristics of the automated assays were examined in terms of both analytical parameters and possible sources of problems that are of potential importance in the operation of an autom ated system in a ro u tin e clinical laboratory. The reproducibility and system stability for the T4, and T3, and T3U assays appeared to be similar to those for other assays, e.g., digoxin, cortisol, and urinary cyclic adenosine m onophosphate acid (AMP) performed with this type of automated instrument.,9,10 The relative intraand inter-assay standard deviations and system stab ility observ ed for these thyroid function assays were comparable to similar assays reported for other autom ated RIA system s4,6,7 and im proved over those commonly reported for manual thyroid function assays. The patient samp le correlatio n s b etw een the serum thyroxine and triiodothyronine concentrations measured with the automated system and the manual procedure were comparable to the correlations reported for other types of automated thyroid function

8 A U T O M A T E D R A D IO IM M U N O A SSA Y F O R T 4, T 3, T 3U, A N D F T I 4 9 assays.4,7 The recovery, dilution, drift, and carry-over studies reported here gave results consistent with routine biochemical thyroid function analysis. In com paring p a tie n t biochem ical thyroid status by using the FTI calculation, the number of automated vs. manual assay discrepant (misclassified) samples was significantly reduced from comparing T3U values exclusively. The two significant outliers that did not correlate by comparison of FTI were analyzed as low T3U by GAMMAFLO with normal T4 by both assays. Clinical evaluation of these two patients did not reveal a clinical picture consistent w ith hypothyroidism. Both were female; no measurements of thyroxine b in d in g g lo b u lin w ere performed. In addition to evaluation of the assay parameters, particular attention was also given to an assessment of the simplicity of operation of the automated instrument. The system was simple to operate, especially for analysts accustomed to presentday computerized clinical chemistry instrumentation. A valuable feature of the instrument for the T4 or T3 assays was that the operator could assess the quality of the standard curve both graphically and numerically before the assay was performed on the entire group of patient samples. The autom ated system provided the num ber of bound counts for the first standard within 15 minutes and established the complete standard curve within approximately 30 minutes from the start of the assay. Sample and control results then follow ed at approxim ately 70 second in terv als. A re c e n t survey of radioimmunoassays indicated that a typical manual thyroid function RIA method required betw een two to four hours for the preparation of the assay and acquisition of a standard curve8 with all samples being processed prior to determination of the standard curve. This instrument has several safety features which are worth noting. The la o 0 e>< 1! n = 5l > T3 ' * Uptake GAMMAFLO (% Uptake) F i g u r e 5. Patient correlation for T3-Uptake assays. Vertical and horizontal lines represent manufacturer s suggested normal reference ranges for combined males and non-pregnant females; Squibb GAMMAFLO:.6 to 40.8 percent uptake and Diagnostic Products 4. to 33.4 percent uptake. belled reagent bottle itself is placed within the pipetting holder, thereby eliminating the manual transfer of radioactive material and the disposal of pipet tips or test tubes X3 o <r o o_ ó < n =51 * T4 Index U * «* G AM M AFLO (T4 Index) FIGURE 6. Patient correlation for F T I (T4 x T3U) for automated vs. manual methods. The indicated reference ranges are from the respective manufacturers suggested ranges based on their determined T4 and T3U normal reference values: Squibb: 1.33 to 4.56 and Diagnostic Products 1.33 to

9 50 V A LD ES A N D U S E T E D (since none are utilized). During the assay all of the radioactive material, representing a very small part of the total fluid volume, is contained within the flow system tubing and the eluant waste container. The container can then be easily capped and d isp o sed as liq u id w aste. O ne suggested improvement would be the addition of an ion exchange resin bed for adsorbing the radioactive w aste and thereby permitting its disposal in a more compact solid form. This is not presently available on the instrum ent evaluated by us. With regard to possible spillage or tubing leakage, the instrument s housing for the pump and the tubing, including separation column, reagents, and samples, is designed to contain the material with easy access for clean-up. Some instrument operation fail-safe mechanisms previously reported9 are: (1) an alarm system which alerts the user of valve timing errors and pre-defined deleterious deviations in sequential sample detection, () automatic counting chamber background counts prior to each run, and (3) sample processing which automatically stops after the last sample in a batch is essayed. There are two areas where we believe this instrument could improve in versatility. In view of the variety of assays adaptable to this system,,9,10 it would be advantageous to have a m ultichannel capability not presently available and, therefore, not limiting assay through-put to a one assay at a time sequential basis. In addition, the availability of a dial-in type of reagent buffer manifold would elim inate the need to change buffer reservoirs physically when required as in changing from the digoxin or cortisol assays to the thyroid function tests reported here. In comparison to manual procedures, this automated RIA system, as we evaluated it, seems to offer the advantages of speed as well as re p ro d u c ib ility and sim plicity of operation. References 1. B r o o k e r, G., T e r a s a k i, W. L., and P r i c e, M. G.: Gam maflow: A com pletely autom ated radioimmunoassay system. Science 194:70-76, B r o o k e r, G. and M u r a d, F.: A utom ated GAMMAFLO radioimmunoassay of urinary cyclic AMP. Clin. Chem. 6: , B r o u g h t o n, P. M. G., B u t t o l p h, M. S., G o w e n l o c k, A. H., N e i l, D. W., and S h e n - t e l b e r y, R. G.: Recommended scheme for the evaluation of instruments for automatic analysis in the clinical biochemistry laboratory. J. Clin. Path. :78-000, E r t i n g s h a u s e n, C., S h a p i r o, S. I., G r e e n, G., and Z b o r o w s k i, G.: Adaptation of a T3-uptake test and of radioimmunoassays for serum digoxin, thyroxine, and triiodothyronine to an autom ated radioimmunoassay system-centria. Clin. Chem. 1: , H e r n e r, A. E. and C l a y t o n - H o p k i n s, J. A.: T rouble-shooting radioligand assays. Clin. Chem. 4: , I s m a i l, A. A., W e s t, P. M., and G o l d i e, D. J-.: The Southmead System : a simple, fullyautom ated continuous-flow system for immunoassays. Clin. Chem. 4: , N y e, L.,A n d e r s o n, M.J., D a w e s, C., L a n d o n, J., and F o r r e s t, G. C.: Automated tests for the assessment of thyroid function. Clin. Chem. Acta 87: , T h e C h e m i s t s G u i d e t o R a d i o a s s a y P r o d u c t s. C lin. C h e m. 4:1-174, V a l d e s, R., J r., S a v o r y, G., B r u n s, D., R e- n o e, B., S a v o r y, J., and W i l l s, M. R.: Performance assessment of the Gammaflo automated radioimmunoassay system by assaying for digoxin. Clin. Chem. 5: , V a l d e s, R., J r., W i l l s, M. R., and S a v o r y, J.: Evaluation of an automated radioimmunoassay for serum cortisol. Ann. Clin. Lab. Sci. 10: , 1980.