Direct and indirect free thyroxine assay methods: theory and practice.
For the diagnosis of thyroid disease, the "free hormone" concept (1) is generally accepted [with exceptions (2-4)] as an appropriate measure. This implies that the supply of thyroxine ([T.sub.4])  or triiodothyronine ([T.sub.3]) into cells is governed by their unbound (free) concentrations in serum rather than the protein-bound fractions. Valid assays measuring the free fraction of [T.sub.4] ([FT.sub.4]) ideally must perform without bias, despite large variations (both absolute and relative) in the concentrations (and affinities) of serum [T.sub.4]-binding proteins found in the population.
No assays measure the actual unbound molecules of [T.sub.4] in serum. Regardless of the probe used, be it a dialysis membrane, a nonspecific binder, or an antibody, unbound [T.sub.4] is abstracted or dissociated in excess of the original unbound hormone in the sample. This does not matter as long as the amount of hormone sampled is so small a fraction of the total (bound and unbound) that the equilibrium between bound and free forms in serum is minimally disturbed.
For a euthyroid subject with serum binding protein concentrations within the appropriate reference intervals, ~0.01-0.02% of the total serum [T.sub.4] is present as [FT.sub.4]. The Mass Action relationship at equilibrium between the two fractions of bound and free hormone for a single protein is:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where [K.sub.equil] is the equilibrium binding constant of the protein for [T.sub.4].
This equation transforms to:
[[FT.sub.4]] = Concentration of [T.sub.4] on occupied sites / [K.sub.equil] X Concentration of unoccupied sites
Consequently, the measured [FT.sub.4] concentration will remain essentially unaltered, provided that the ratio of occupied/unoccupied sites holds virtually constant. Hence, measurements of [FT.sub.4] are valid only with minimum dissociation of bound [T.sub.4] from the serum proteins. From the known equilibrium constants for the three serum binding proteins, thyroxine-binding globulin (TBG), transthyretin, and albumin (5, 6), a maximum of 5% of the protein-bound [T.sub.4] in serum can be dissociated into the free form, but preferably much less than this (7).
Any valid assay must encompass the whole physiologic range of serum binding protein concentrations. This can be defined as the ideal "window of validity' for any [FT.sub.4] assay. Within this window, no serum presented to the assay is affected (with regard to disturbance of the bound-free equilibrium for [T.sub.4]) so as to be significantly influenced by either the [T.sub.4] sampling process or by procedures that might otherwise affect the equilibrium. Minimization of disturbance is especially difficult for sera with very low binding capacities for [T.sub.4] (e.g., with low concentrations of TBG and/or albumin). When operating outside their windows, assays no longer measure [FT.sub.4]. Various methods adhere to the above requirement to different extents, and even within any given common technology, individual assays may differ detectably in their compliance. Hence, uniquely in immunoassay technology, the mode of operation and window of validity of any given [FT.sub.4] assay (or its equivalent) are individually defined, so that subtle differences arise in the assay's performance when compared with any other. Several factors cause this. These do not merely concern arbitrary judgments by commercial manufacturers, or perhaps lack of awareness, as to how extensively the wide variations in serum in the human population should be examined during development of their assays. They also include differences in concentration, specificity, and affinity for [T.sub.4] of any antibodies used; the extent of serum dilution by assay ingredients; the choice of assay reagents such as buffers to mimic physiologic conditions; the choice and concentration of analog or competitive binding reagent; and the time and temperature of the assay incubation process (7-14).
Given complexity of this magnitude and the failure of many producers to fully grasp its implications, it is unsurprising that, even after 20 years of development, individual [FT.sub.4] methods still vary considerably in performance (i.e., their windows of validity fall short of the ideal). However, regardless of these differences, routine laboratories can perform useful investigations to test the validity of each assay method. This review, in choosing the most commonly used methods for the measurement of [FT.sub.4] in serum, aims to reevaluate their validity and to suggest ways in which the limits of this validity can be simply tested. Individual named assays will not be examined in detail. There are too many to be evaluated in a limited space, and the continual appearance and replacement of commercial assays would render any review rapidly obsolete.
The following analysis of methods for measuring [FT.sub.4] includes an examination of the [T.sub.3] uptake test [and thus the indirect free thyroxine index (FTI) approach], and explains why shortcomings in performance can occur (9, 15-18). Direct one- and two-step immunoassays are then considered, comparing these in concept and performance with assays ranging from the FTI to the so-called "gold-standard" methods of equilibrium dialysis and ultrafiltration.
The direct one-step assays have recently been criticized as both failing to correctly measure [FT.sub.4] concentrations in protein-free and other artificial solutions and to obey classic tests for recovery of analyte (19-24). Many of these experiments have, however, been conducted outside the window of validity of the assays. For example, if the antibody in valid [FT.sub.4] assays can be allowed to bind 0.01-1% of the total hormone in a serum sample, recovery estimations, in the classic sense, have no meaning. It is not surprising, therefore, that studies of artificial solutions containing serum [T.sub.4] binding proteins and [T.sub.4] showed apparent correlations of [FT.sub.4] with binding protein concentrations (19-24). These tests were also performed under conditions in which the equilibrium between bound and free hormone was sufficiently perturbed to disobey the essential criterion for a valid assay. Such experiments, in which (unlike the case with equilibrium dialysis or ultrafiltration) the serum binding proteins, their [T.sub.4] load, and the antibody probe interact in solution, teach the need for caution in devising suitable tests for assay validity.
IMPORTANT FACTORS AFFECTING [FT.sub.4] MEASUREMENTS
As emphasized earlier, the primary rule for valid measurement of [FT.sub.4] in any assay is that, by the assay process, serum protein-bound [T.sub.4] is minimally displaced into the free phase, with insignificant disturbance of the endogenous bound-free equilibrium for [T.sub.4] in the sample (7, 9). The permissible amount of hormone so displaced can be many times that of the concentration of endogenous [FT.sub.4] (7, 9). Disturbances of the equilibrium can take several forms. [T.sub.4] can be sequestrated, through the free fraction, onto the binding sites of a probing antibody. Simultaneously, the bound hormone is further partially dissociated simply by dilution of the serum when assay ingredients for [FT.sub.4] measurement are added. In gold-standard methods, such as equilibrium dialysis, dilution is the key function to control, because the serum is first dialyzed before the separated dialysate is assayed for [T.sub.4] by a sensitive total-hormone assay (25-27). Conversely, in direct one- or two-step immunoassays for [FT.sub.4], both dilution of serum and simultaneous sequestration of [T.sub.4] by the antibody probe are equally relevant. These influences are, in the latter case, approximately additive. If the antibody sequestrates 1% of the total [T.sub.4] from a sample containing 0.02% free in the original serum, this is equivalent to a dilution of 50-fold. If the addition of assay ingredients to the sample dilutes it by 10-fold, then the overall effect is approximately equivalent to a 60-fold dilution. If, however, the original serum sample was first diluted 2-fold (e.g., to conduct a dilution test), then the fixed amount of antibody will remove ~2% of the total hormone from the serum proteins and this, combined with the 10-fold dilution of the assay process itself, will approximate to a 110-fold dilution. Equivalent effects occur in a serum naturally containing one-half the normal binding capacity for [T.sub.4] (i.e., sera with lower endogenous binding protein concentrations should act in a given assay as if a serum with concentrations within the reference intervals had been diluted appropriately).
In some [FT.sub.4] immunoassays, one of the assay ingredients may be albumin, which was initially added to modulate the residual binding of analog to the serum [T.sub.4]-binding proteins (7) but was subsequently used to mitigate the effects of substances such as nonesterified fatty acids (NEFAs) on [T.sub.4] binding to the serum binding proteins. However, exogenous albumin may also distort (lower) true serum [FT.sub.4] values by its own sequestration of [T.sub.4] from the serum equilibrium system. (A detailed discussion will be presented later in the text.)
For the earlier [T.sub.3] uptake test used in determining FTI, two simultaneous influences are at work. One is the potential of the competing binder (often nonspecific), used to bind the excess [T.sub.3] not bound to the serum binding proteins, to also bind [T.sub.4] (18). The other is the potential for the high concentration of the [T.sub.3] label, necessary to saturate the unoccupied binding sites, to dissociate some of the bound [T.sub.4] from its bound sites on the serum proteins by direct competition. From Mass Action equations, relatively little of the bound [T.sub.4] need be dissociated before there is unacceptable distortion of the FTI result away from a valid [FT.sub.4] measurement.
Finally, factors either endogenous in the sample in vivo or developed in vitro during assay incubation may interfere with [T.sub.4] binding in serum, again displacing enough bound [T.sub.4] to seriously affect (increase) the [FT.sub.4] or FTI measurement. This is especially noticeable when certain drugs or other substances are present that compete with [T.sub.4] for some of its serum protein binding sites, or where similarly competing NEFAs are produced in vitro in heparinized patients (28-33).
All of these factors must be evaluated and their effects recognized in the development of valid assays for [FT.sub.4]. Failure to do so, especially with regard to antibody sequestration of [T.sub.4] and the extent of intrinsic serum dilution by an assay, can compromise the validity of the measurements, so that [FT.sub.4] values may correlate with one or more of the serum binding protein concentrations. The same phenomenon will occur with the [T.sub.3] uptake test if the amount of label and the effect of the competing binder unduly disturb the natural bound-free equilibrium of the serum proteins.
Early Approaches to [FT.sub.4] Measurement
The FTI approach (34, 35) provided the first convenient estimation of [FT.sub.4] in serum in the routine laboratory. This method combines two measurements: total [T.sub.4] by immunoassay and [T.sub.3] uptake. From the basic Mass Action equation described earlier, total [T.sub.4] closely approximates the serum protein-bound [T.sub.4] (occupied sites), and [T.sub.3] uptake represents the concentration of unoccupied sites. FTI, as an index of [FT.sub.4], is therefore expressed as the product of the two tests/100. The values obtained do not, of course, represent actual [FT.sub.4] concentrations (34).
The total [T.sub.4] measurement of the FTI is perfectly valid. However, the commonly held interpretation of the way in which the [T.sub.3] uptake test works (34) is probably incorrect. This holds that the [T.sub.3] added merely occupies the vacant binding sites on the serum [T.sub.4]-binding proteins, the excess spilling over for collection by a competing binding agent. Scrutiny of the Law of Mass Action casts doubt on this simple interpretation.
Serum contains two classes of [T.sub.4] binding proteins: TBG, as a low-concentration, high-affinity protein; and transthyretin and albumin as high-concentration, low-affinity binders. For a normal euthyroid serum, the TBG concentration is ~2000-fold lower than the combined concentrations of transthyretin and albumin, but TBG binds ~60% of the total [T.sub.4] present (5, 6). Whereas most of the TBG binding sites are occupied, <1% of the transthyretin and albumin sites bind [T.sub.4]. When excess [T.sub.3] is added in the [T.sub.3] uptake test, enough must be added to both occupy high- and low-affinity binding proteins and provide sufficient excess for sequestration by the added binding probe. This creates a dilemma. If enough [T.sub.3] must be added to fill the large excess of high-concentration, low-affinity binding sites (virtually all of which are unoccupied), then this concentration will be high enough to disturb [T.sub.4] binding to TBG. The affinity of TBG for [T.sub.3] is only ~10% that for [T.sub.4] (36-38), but the quantity of [T.sub.3] necessary for the assay will outweigh this smaller affinity, allowing [T.sub.3] to significantly displace [T.sub.4] from the TBG binding sites. The effect of the fixed amount of added [T.sub.3] on the [T.sub.4] displacement phenomenon will be proportionately smaller for high-TBG sera and proportionately greater for low-TBG sera. The [T.sub.3] uptake test (through the FTI) thus inevitably displays a TBG dependency (14-16,18). Thus, because the added labeled [T.sub.3] is preferentially bound to the large excess of unoccupied transthyretin and albumin sites, the [T.sub.3] uptake test should also produce falsely low FTI results in nonthyroidal illness (NTI) where the concentrations of such proteins are often reduced (39-42).
The nonspecific binders used in many [T.sub.3] uptake assays (18) can also create additional problems by directly sequestrating some [T.sub.4] as well as [T.sub.3]. With such a variety of effects (14,16-18), it is not surprising that, for example, the imperfect normalization of, e.g., high-TBG late-pregnancy sera into the mid- or high-normal range for FTI should have been accepted historically as the correct placement, when more accurate measurements lacking TBG distortions place such sera lower in the reference interval (43-45).
The FTI approach is therefore one in which the rules for valid [FT.sub.4] assays are not met. It produces a hybrid result, with a very narrow window of validity, between a total [T.sub.4] measurement and a true [FT.sub.4] measurement. The positioning of a given FTI in the scale between these two extremes depends on the conditions defining the [T.sub.3]-uptake test (45). These include the concentration of [T.sub.3] used relative to the volume of serum tested, the degree of dilution of serum in the assay conditions, and the amount and specificity of the competing binder used to sequestrate the excess [T.sub.3]. Assays using more specific secondary binders that do not incidentally sequestrate [T.sub.4] from serum should be superior to those that use nonspecific binders.
The routine laboratory can probe the relationship of a given FTI with TBG concentrations in serum (which should be carried out only with a panel of euthyroid sera having a wide range of TBG concentrations from otherwise healthy subjects, and excluding either pregnant or nonthyroidally ill subjects). It can also place the FTI measurements along the arbitrary scale from total [T.sub.4] to [FT.sub.4] (46). One need only measure total [T.sub.4], FTI, and [FT.sub.4] by selected methods on a panel of ~150-200 routine (not severely ill or pregnant) subjects (9), including some hypo- and hyperthyroid patients, to encompass virtually the whole range of expected values for each marker. It is assumed that each assay method has approximately the same imprecision. Any pair of the three markers is now correlated. If the FTI is acting as a hybrid assay, imperfectly correcting for the natural variation in serum binding protein concentrations in the panel, then it will correlate more closely with total [T.sub.4] than will the [FT.sub.4] assay (9). The extent to which the correlations differ is a rough measure of how effectively the FTI approach corrects for binding-protein differences. If changing from FTI to an [FT.sub.4] assay is contemplated, then several [FT.sub.4] assays could be used as comparators. In general, given equivalent imprecision, the smaller the correlation coefficient with total [T.sub.4], the more likely is it that the assay in question is correcting completely for serum binding protein variations.
In addition, a valid [FT.sub.4] assay should show better discrimination than FTI at the euthyroid-hyperthyroid borderline of the reference interval (9, 46). Performance at the hyperhyroid borderline may not be so clearly altered (9, 46).
THYROXINE ([T.sub.4]) UPTAKE (T-UPTAKE) ASSAYS
In this method, labeled [T.sub.4] (or a suitable analog) is added in large excess to serum (47). The excess hormone (or analog) fills the hitherto unoccupied protein [T.sub.4]-binding sites and spills over into a free fraction, which can be measured. Just as for [T.sub.3] uptake methods, [T.sub.4] uptake techniques (47) may suffer from similar distortions arising from the very different concentrations and binding affinities of TBG on the one hand vs transthyretin and albumin on the other (see above). Before addition of additional hormone, TBG sites are already occupied by endogenous [T.sub.4] to a large extent, whereas transthyretin and albumin sites are not. Much of the added hormone is thus bound to transthyretin and albumin. This means that the distribution of added hormone is more sensitive to differences, both relative and absolute, in the concentrations of these binding proteins. A decrease in such concentrations relative to TBG, especially in NTI, may strongly affect the [T.sub.4]-uptake value, giving, in turn, low index values (48). These problems, as with [T.sub.3] uptake, again produce a hybrid result, with a truncated window of validity, in which serum [T.sub.4]-binding protein effects distort the values. The only advantage over the [T.sub.3] uptake test is that when labeled [T.sub.4] is used, the same hormone is being used to assess unoccupied sites, with the same affinities as the hormone already occupying the serum binding protein sites. However, if a labeled [T.sub.4] analog is used instead (T-uptake), its affinity for TBG (relative to [T.sub.4]) may be attenuated more than for transthyretin and albumin. This will exaggerate the influence of transthyretin and albumin even further and may lead to expected falsely low results in NTI, where concentrations of these proteins usually are low.
Direct [FT.sub.4] Immunoassays
In the period 1978-1980, challenges arose to the hegemony of FTI, which fell increasingly under suspicion of inadequate performance. These included a variety of so-called kinetic [FT.sub.4] assays (49, 50), followed by a two-step approach (51). The former assays had only a relatively brief existence and are now obsolete. However, the two-step approach to [FT.sub.4] was basically valid (51) and was a major advance over what was then available commercially. In this test, serum is first incubated with a small quantity of immobilized [T.sub.4]-specific antibody, which takes up a very small quantity of [T.sub.4] from the sample, minimally disturbing the original serum bound-free equilibrium. After equilibrium is reached, the serum and antibody are separated (51). The immobilized antibody containing bound [T.sub.4] is then washed repeatedly to remove traces of serum. It is then incubated with a labeled probe (e.g., a suitable [T.sub.4] derivative or, initially, [sup.125]I-labeled [T.sub.4]), and the occupancy of antibody binding sites is measured as for a typical immunoassay.
The two-step assay no doubt is independent, by definition, of the influence of the serum proteins and their bound [T.sub.4]. There are, however, potential problems that must be addressed. The first is the need to remove rigorously all traces of serum proteins before the second incubation is begun. Even minor amounts of binding proteins could sequestrate the labeled probe because of their high affinity. Modern two-step assays avoid this problem by use of [T.sub.4] conjugated with macromolecules (e.g., enzymes). Such conjugates usually have negligible affinity for serum protein binding sites. A second problem is the possibility of "back displacement" of [T.sub.4] bound in the first incubation by competition with labeled probe in the second, and by the resulting nonequilibrium conditions for [T.sub.4] binding (i.e., the phase of serum and its remaining [T.sub.4] has been washed away). Again, macromolecular [T.sub.4] conjugates can minimize the first effect because usually the conjugate will bind much less avidly to the antibody than does [T.sub.4]. The use of [T.sub.3] conjugates may reduce competition even further. The potential for re-equilibration of antibody-bound [T.sub.4] remains however, and suitably avid antibodies should be used to minimize this effect. Short (nonequilibrium) second incubations can also help.
ONE-STEP ANALOG IMMUNOASSAYS
To avoid the inconvenience of using two separate processes, with their potential for greater overall imprecision, Dr. T.A. Wilkins and I invented and developed the one-step analog technique for the measurement of [FT.sub.4] (8, 9, 52) and F[T.sub.3] (53). In these assays, the labeled [T.sub.4] ([T.sub.3]) then used as a competing species for antibody binding was replaced by a chemically modified [T.sub.4] ([T.sub.3]) analog. This had the property of binding avidly to a [T.sub.4] ([T.sub.3])specific antibody, but much less avidly than the native hormone to the serum binding proteins (7-9, 54). The aim of the system was to improve the convenience of two-step approaches while maintaining validity.
Accordingly, one-step approaches made redundant the initial separation of serum binding proteins required by the two-step approach (51) and superficially resembled the corresponding total-hormone assays. Early analog methods, although they performed adequately in correcting for the physiologically encountered differences in TBG and transthyretin concentrations in serum, were correlated with serum albumin concentrations (7, 55-58). Although the assays were at the forefront compared with anything else then commercially available, they were criticized for imperfect performance (55-60). Vehement controversy between us and others (2-4, 7, 59-63) as to the exact working and validity of such assays regrettably generated far more heat than light and led to hesitation in some countries (especially the US) in adopting the methodology. The critics, however, paid little attention to the fact that the alternative FTI methods, the most frequently used assays at the time, had greater shortcomings in the same and other areas. As an example of this critical misinterpretation, the correlation of late-pregnancy sera that produced low-normal analog [FT.sub.4] values with serum albumin concentrations was alleged to demonstrate the invalidity of these assays (55). It is now known that this correlation is merely a physiologic one and not an assay shortcoming (45). Other studies alleged, wrongly, that the assays were merely measuring the "albumin-bound" fraction of [T.sub.4], and not [FT.sub.4] (57). Practical experience in the development of the modern one-step assays lacking interferences also showed that the stringent criteria laid down by theorists (2-4, 59, 60) for the permitted amounts and affinities of antibodies in valid [FT.sub.4] assays could be considerably relaxed and still give acceptable dose--response values (64, 65).
At first (8, 9), the enforced use of radioactive labeling probes limited analog immunoassay design. Assay development was constrained by the maximum specific activity of the label compatible with the analog's chemical stability. Detection of adequate amounts of signal for convenient analysis required the antibody to sequestrate slightly >1% of the total hormone in euthyroid sera with reference binding protein concentrations (7-9), still well within the window of validity for the assay (7-9). Albumin was also added to modulate the biasing effects of variable serum albumin concentrations (7).
Since then, improvements in the performance of analog [FT.sub.4] and F[T.sub.3] assays (64, 65) have eliminated interactions of the labeled probe with any of the common serum binding proteins. Newer nonradioactive labels, with higher detection sensitivity, use smaller amounts of antibody, consequently extracting a lower percentage of hormone (64, 65). This gives three advantages. The first advantage is that when measuring sera with abnormal [T.sub.4]-binding capacities, intrinsic bias is minimized. The second advantage (65) is that, when they lack added albumin, assays are more robust to progressive serum dilution (a useful technique, although sometimes overinterpreted as a test for assay validity and absence of bias). The third advantage is that [FT.sub.4] results no longer significantly correlate with serum binding protein concentrations, even albumin (64-66).
Comparison of the performance of the newer and original assays confirmed earlier forecasts (7) that the need for improvement lay in the abolition of biasing effects attributable to albumin. This comprises the hypoalbuminemia of severe NTI (still a controversial area in regard to the value of thyroid function tests) and the rare conditions of analbuminemia, familial dysalbuminemic hyperthyroxinemia, and the presence in serum of strongly binding [T.sub.4] or [T.sub.3] autoantibodies. The first three have been addressed with good analog one-step assays (67, 68), and all, including the vast majority of sera containing autoantibodies, have been addressed successfully in the newer immunometric assays, where the antibody rather than the analog is labeled (64, 65).
With modern one-step assays, therefore, comparing new methods with earlier ones is more difficult than comparing an FTI with any [FT.sub.4] method. This is because, as stated above, the possible improvements in diagnostic performance become increasingly marginal as the learning curve in development progresses. For the routine laboratory, the most relevant test of the validity of any [FT.sub.4] assay is to correlate, in euthyroid sera with wide variations in binding protein concentrations, the [FT.sub.4] value with each of the proteins TBG, transthyretin, and albumin. Once again, subjects receiving drugs such as dilantin or aspirin, and nonthyroidally ill and pregnant subjects should not be included in the correlations because these may have altered physiology dependent on their condition. Specificity and sensitivity measurements are also recommended for discrimination of hypo- and hyperthyroidism from the reference interval.
Dilution tests on sera are valuable, but require restraint. The useful limits of dilution experiments can be calculated from the affinity constants and concentrations of each of the serum [T.sub.4] binding proteins, for which the binding potential (BP) is given as the product of the two parameters ([Sigma][K.sub.1,2,3][P.sub.1,2,3]). For a normal euthyroid serum, BP is ~6000 (4000 from TBG,1300 from transthyretin, and 700 from albumin). Sera such as those in NTI could have a BP value ~15-20% of this (i.e., ~900-1200). Therefore, dilution tests on a normal serum need only be pursued out to a dilution (over and above the dilution of serum intrinsic in the assay itself) of no more than five- to sixfold. If the assay in question is sufficiently (not necessarily perfectly) robust out to this dilution factor, then it is likely that sera with BP values of 900-1200 are being measured without undue bias. Sera with BP values lower than this are relatively rare. Dilution of sera with low BP values is unnecessary except as an academic exercise to probe the limits of validity of the assay. The methodology is then being stretched into nonphysiologic regions that may exceed its window of validity, and false indications of assay failure may ensue. It may well be that an assay is designed to encompass, without undue bias, the BP values for the vast majority of sera encountered, but can rapidly fail outside this range.
Comparison of results with those from a gold-standard method such as equilibrium dialysis or ultrafiltration is also valuable, if such facilities are available. However, careful choices of sera should be made, and nonthyroidally ill subjects should not generally be included, nor should serum from patients known to have taken drugs that can affect the binding of [T.sub.4] to the serum binding proteins (see below).
IMMUNOMETRIC ONE-STEP ANALOG METHODS
More recently, further advances have been made in one-step [FT.sub.4] analog technology, with two key innovations (64, 65). The first innovation involved transferring the detection label from the analog itself to a specific [T.sub.4] binding monoclonal antibody and immobilizing the analog as a complex insoluble matrix. In the second innovation, a cross-reacting hormone ([T.sub.3]) was used rather than [T.sub.4] itself for binding to antibody as the basis of insoluble matrix formation (insoluble [T.sub.3] matrix) (64, 65). This invention confronted the outstanding problem of how to measure [FT.sub.4] in sera containing avid autoantibodies against [T.sub.4] or [T.sub.3], as well as finally preventing the interference of any other serum binding protein in the assay (64-66). In this assay, the affinity of the insoluble [T.sub.3]-matrix for the antibody is manyfold less than for [FT.sub.4]. Thus, so that the relative rates of binding of the insoluble [T.sub.3] matrix and [FT.sub.4] to the labeled antibody are roughly equivalent, the concentration of this matrix must be correspondingly manyfold higher than [FT.sub.4] (64, 65). Hence, occupation of the immobilized [T.sub.3] matrix by a small quantity of the antibody probe is negligible. This allows the gross excess of unoccupied insoluble [T.sub.3] matrix binding sites to bind large amounts of extraneous autoantibodies without significantly affecting the binding ability of the labeled antibody probe (64, 65, 69, 70). The method has been shown to accommodate the vast majority of such sera within the correct range for their underlying functional state (69, 70), unlike labeled-analog methods, which are still often susceptible to interference (71).
In addition, it was realized that the basic mode of action of direct [FT.sub.4] immunoassays, using avid antibodies, could be expressed as a simple competition for binding of [FT.sub.4] and analog (or insoluble analog-matrix) until an apparent equilibrium is reached (64, 65). The small, extremely slow dissociation of antibody-bound analyte and analog thus greatly simplified the original complex arguments (7) that sought to explain the assay system (64, 65), permitting the amounts and affinities of antibody and analog to be more variable than the original, more rigid theory allowed (2-4, 7, 59, 60).
ALBUMIN IN THE ASSAY INGREDIENTS IN ONE-STEP ASSAYS
Initially, extra albumin was provided in the assay ingredients of the first one-step assays to ameliorate the biasing effects of residual binding of the analog tracer to endogenous albumin in the serum sample (7). This inevitably inserted an additional bias. The smaller the binding protein concentrations in any given serum, the greater was the influence of the added albumin on the assay through its own [T.sub.4] sequestration. However, direct experiments, using added albumin, showed that (a) the binding properties of added albumin were variable and (b) the bias was important only when serum albumin concentrations (but not those of TBG or transthyretin) differed significantly from the norm (7, 58). Accordingly, in these assays, [FT.sub.4] measurements correlated only with serum albumin concentrations (7, 66). Even then, the effects were not marked until albumin concentrations had fallen well below the reference interval for that protein [e.g., in conditions ranging from severe NTI to analbuminemia (7, 58)]. Observed bias within the reference interval for albumin concentrations did not have significant consequences for diagnostic discrimination (66).
Over the next 20 years, continual improvements in the properties of the analogs eventually led to the abolition of their binding to serum proteins (64, 65). Accordingly, there was no need to add albumin expressly to counteract bias from residual binding effects. However, another problem manifested itself, which extra albumin in the assay ingredients helped to ameliorate. In patients receiving heparin injections, blood lipoprotein lipase is activated (72, 73). When blood is withdrawn from such patients, NEFAs are produced in vitro, often to very high concentrations (74-78). The NEFAs in turn are capable of displacing substantial quantities of bound [T.sub.4] from the serum protein [T.sub.4]-binding sites and can increase the measured [FT.sub.4] considerably, often grossly out of the reference interval (30, 31, 74-78). Albumin in the assay ingredients opposed this artificial increase in [FT.sub.4] by "mopping up" the NEFAs produced, thereby restoring most values to normality (30, 31). Thus, assays containing albumin can be used with heparinized subjects (with caution).
In severe NTI, there are additional problems. Initially, in the days of FTI supremacy, it was believed that without overt primary thyroid dysfunction, such patients were essentially euthyroid, although there were considerable decreases in the concentration of total [T.sub.4] into the hypothyroid range, not wholly corrected by FTI (40-42, 79, 80).
Gold-standard methods for [FT.sub.4] measurement variously gave results below, within, and above the euthyroid reference interval (27, 41, 42, 81-85). Simultaneously, thyrotropin (TSH) concentrations could also be low, normal, or occasionally above normal (86-89), although not necessarily in harmony with the [FT.sub.4] values. It was postulated that in NTI, the serum contained nondialyzable substances that could bind competitively to the [T.sub.4]-binding proteins, displacing bound [T.sub.4] into the free phase (90-92). This explanation rationalized how the gold-standard [FT.sub.4] measurements could be so out of line with total [T.sub.4] values, even when the serum binding protein concentrations (especially transthyretin and albumin) were also severely decreased. Nevertheless, [FT.sub.4] estimates straddling the reference interval also create problems in discriminating true primary hypo- and hyperthyroidism in subjects with NTI from subjects with no apparent thyroid dysfunction.
It was, no doubt, hoped that when new, more convenient assays were available, these anomalies would disappear. However, over the past 20 years, the exact status of the NTI subject with regard to [FT.sub.4] in serum has remained as controversial as ever and certainly cannot be conceived as "steady-state" with regard to thyroid function indicators (93). In NTI, overlaying any general level of thyroid function, there seem to be continuous (independent) ebbs and surges in the production and retention of substances either directly or indirectly related to [FT.sub.4] as the disease progresses (81-92). These can include serum binding proteins (94); interfering substances, including administered drugs (32, 33); TSH (90-92); and probably production of [T.sub.4] and its conversion to [T.sub.3] (95, 96). The true thyroidal state of the subject can thus be obscured by relatively short-lived changes in measured indicators, which may, at random, show up in single samples taken from a panel of NTI patients. This highlights the danger of diagnosing NTI patients by single samples, making it difficult to compare assay performance if the panel of sera inevitably contains examples of these random events. Indeed, because of the tendency of [FT.sub.4] and TSH concentrations independently to enter and leave the euthyroid reference interval, it is recommended that both markers be measured in NTI and that careful monitoring is done to rule out effects that might alter the concentration of either marker, e.g., administration of albumin-binding drugs (66, 97). Only in this way can we assess the underlying integrated level of thyroid function and discount transient anomalies. The difficulties in diagnosing hypothyroidism accompanying NTI have been vividly described (98).
Most studies comparing [FT.sub.4] assay performance in NTI have not taken this complexity fully into account and are generally poorly designed. Assays with biases attributable to added albumin have often been criticized for giving below-normal values when TSH is presumably within the reference interval (31, 32, 96). However, because gold-standard assays lacking added albumin can equally give diagnostically meaningless above-normal values, misplaced confidence in their validity has often led to support for assays that show the same effect (94). Here there is a logical inconsistency. If albumin-containing [FT.sub.4] assays can read too low and TSH is, in a random NTI sample, transiently high, a misdiagnosis of hypothyroidism could occur. Conversely, if albumin-free assays give above-normal values when TSH is transiently low, a misdiagnosis of hyperthyroidism could equally be made. Neither methodology thus can be completely trusted in this situation. Additionally, albumin-containing assays may mitigate the transient effects on [FT.sub.4] of albumin-binding drugs in serum and depict the underlying thyroidal function more exactly.
Finally, the potential effects of assay incubation on the concentration of interfering substances in serum should not be underestimated. In an "albumin-free" assay using a serum whose [T.sub.4]-binding sites are already saturated with other molecules, the [FT.sub.4] measurement is exquisitely sensitive to any further production of interference in vitro. Published studies show that [FT.sub.4] values are increased when NEFA concentrations are >2.5 mmol/L (78, 99-101). The rate of further increase of [FT.sub.4] on increases in NEFAs is ~7-10 pmol/L per mmol/L additional NEFAs (7, 30, 78). Therefore, an increase in NEFAs (or the combination of some other interferent + NEFAs) of 0.2 mmol/L will cause an (in vitro-dependent) increase in measured [FT.sub.4] of ~1.5-2 pmol/L. Albumin-containing assays can at least partially mitigate this effect (30). The amount of exogenous albumin in the assay buffers that can be used to minimize interference in [FT.sub.4] assays depends very much on its properties of [T.sub.4] binding as well as its concentration. It is well known that commercial sources of albumin can vary markedly in their [T.sub.4] binding characteristics (58). Such sources are properly termed preparations rather than well-defined purified compounds because denaturation of albumin can occur to various degrees during production. However, a calculation can be made of how large a BP for added albumin will distort the [FT.sub.4] values of most sera to an acceptably minimal extent relative to the assay calibrators. This would approximate, at most, to 10-25% of that found endogenously in a serum with binding protein concentrations within the reference intervals. Thus, in an assay using a 10-[micro]L sample and 100 [micro]L of reagent, the bovine serum albumin content of the reagent should be <1.2 g/L.
Hence, [FT.sub.4] values are not grossly affected until endogenous albumin concentrations are less than half-normal and TBG concentrations are also substantially diminished. Albumin preparations with weaker [T.sub.4] binding affinities can be used in greater quantities because the smaller affinity offsets the effects of higher concentration in producing a suitable BP factor. Supplemental data on this topic, in the form of an Appendix, are available in the on-line version of this journal (http://www.clinchem.org/ content/vol47/issue8/).
Selecting assays that either do or do not contain albumin in their buffers appears to be a trade-off of intrinsic bias in [FT.sub.4] values downward in NTI (and in all other situations where the binding capacity is reduced), caused by the added albumin, against increased sensitivity of an albumin-free assay to extraneous in vivo and in vitro effects. Depending on their experience with discrimination of thyroidal disease in NTI subjects, assay users must decide for themselves which to use.
A carefully controlled study on large numbers of subjects, with a defined and graded severity of systemic illness, is urgently needed, with full knowledge of medication taken and with multiple sampling, in which outcomes are known and where substantial numbers of NTI patients with primary hypo- and hyperthyroidism are also included. Such a study, using an albumin-containing and an albumin-free assay for comparison, together with TSH measurements, would more usefully evaluate the sensitivity and specificity of each assay in diagnosing thyroid disease in these conditions. In this way, one could offset the misdiagnosis frequency from the inevitable biases in the albumin-containing assay against the misdiagnosis frequency in albumin-free assays.
It is unfortunate that albumin is the only useful (and inexpensive) protein able to mitigate the effects of interfering substances on [FT.sub.4] values in heparinized, drug-affected, or nonthyroidally ill patients. However, chemical modification of albumin may greatly attenuate or destroy its binding of [T.sub.4] without destroying its ability to bind other substances, such as NEFAs (102). The binding sites for NEFAs and other substances are manyfold more than for [T.sub.4]. Judicious protein modification might produce a substance still able to quench interference without causing significant bias through extra [T.sub.4] binding. This might yield an assay giving superior discrimination in NTI through nullification of the influence of interfering substances while not otherwise suffering from bias from albumin addition.
The performance of convenient assays for [FT.sub.4] has improved greatly since the first attempts (FTI) in 1965. Index methods were accepted until it became clear that they were still strongly influenced by TBG. In the meantime, certain "truths' had come to be accepted by many workers, not the least that in late pregnancy, [FT.sub.4] results should lie in the middle of or in the upper regions of the euthyroid reference intervals. Better assays, correcting more completely for TBG concentrations, forced a sometimes painful period of readjustment in this and other areas, which has taken 20 years or more to reconcile in some regions of the world. By now, the better assays for [FT.sub.4] have eliminated the biasing effects that arose in the early assays from residual binding of labels to serum albumin. The most recent improvements have also largely addressed the remaining problem of [T.sub.4] autoantibodies in serum. The main problem outstanding in using [FT.sub.4] assays is that of diagnosis in NTI. This is a problem that lies more in the vagaries of patient physiology than in any hope of a final, simple, and definitive answer in an ideal assay. The scope for further improvement in [FT.sub.4] assay methodology is very limited. Most of the residual problems (albumin dependency, dilution performance) have been addressed, and a final decision can be taken by diagnosticians, not merely whether to terminate their use of FTI, with all its shortcomings, but which direct [FT.sub.4] assay to use, based on simple comparisons of each assay's diagnostic properties with those of other assays.
Received November S, 2000; accepted May 15, 2001.
(1.) Robbins J, Rall JE. The iodine-containing hormones in blood. In: Gray CH, James VHT, eds. Hormones in blood. London: Academic Press, 1979:576-688.
(2.) Ekins R. Measurement of free hormones in blood [Review]. Endocr Rev 1990;11:5-46.
(3.) Ekins R. The free hormone hypothesis and measurement of free hormones [Editorial]. Clin Chem 1992;38:1289-93.
(4.) Ekins R. The free hormone hypothesis and measurement of free hormones [Letter]. Clin Chem 1993;39:1343-4.
(5.) Tabachnick M. Thyroxine-protein interactions. IV. Thermodynamic values of thyroxine with human serum albumin. J Biol Chem 1967; 242:1646-50.
(6.) Prince H, Ramsden DB. A new theoretical description of the binding of thyroid hormones by serum proteins. Clin Endocrinol 1977;7:307-24.
(7.) Wilkins TA, Midgley JEM, Barron N. Comprehensive study of a thyroxin-analog-based assay for free thyroxin ("Amerlex [FT.sub.4]"). Clin Chem 1985;31:1644-53.
(8.) Midgley JEM, Wilkins TA, Inventors. A method for determining the free portions of substances in biological fluids. European Patent No. 0026103, 1981.
(9.) Midgley JEM, Wilkins TA. The direct estimation of free hormones by a simple equilibrium radioimmunoassay. Amersham, UK: Amersham International Ltd, 1982.
(10.) Nelson JC, Weiss RM. The effect of serum dilution on free thyroxine ([FT.sub.4]) concentrations in the low-[T.sub.4] syndrome of nonthyroidal illness. J Clin Endocrinol Metab 1985;61:239-46.
(11.) Midgley JEM, Wilkins TA, Moon CR. Validity of analog free thyroxin immunoassays. Part II [Opinion]. Clin Chem 1987;33: 2145-8.
(12.) Ross HA, Benraad TJ. Is free thyroxine accurately measured at room temperature? Clin Chem 1992;38:880-7.
(13.) van der Sluijs Veer G, Vermes I, Bonte HA, Hoorn RKJ. Temperature effects of free thyroxine measurements: analytical and clinical consequences. Clin Chem 1992;38:1327-31.
(14.) Docter R, van Toor H, Krenning EP, de Jong M, Hennemann G. Free thyroxine assessed with three assays in sera of patients with nonthyroidal illness and of subjects with abnormal concentrations of thyroxine-binding proteins. Clin Chem 1993;39: 1668-74.
(15.) Konno N. Serum thyrotropin response to thyrotropin-releasing hormone and free thyroid hormone indices in patients with familial thyroxine-binding globulin deficiency. Endocrinol Jpn 1976; 23:313-7.
(16.) Roosdorp N, Joustra N. A numerical comparison of the use of [T.sub.3]-uptake values and of TBG levels for the estimation of free thyroxine in serum. Clin Chim Acta 1979;98:27-33.
(17.) Wilke TJ. A challenge of several concepts of free thyroxin index for assessing thyroid status in patients with altered thyroid-binding protein capacity. Clin Chem 1983;29:56-9.
(18.) Nelson JC, Tomei RT. Dependence of the thyroxin/thyroxin-binding globulin (TBG) ratio and the free thyroxin index on TBG concentrations. Clin Chem 1989;35:541-4.
(19.) Nelson JC, Tomei RT. Direct determination of free thyroxine in undiluted serum by equilibrium dialysis/radioimmunoassay. Clin Chem 1988;34:1737-44.
(20.) Nelson JC, Wilcox RB. Protein bound [T.sub.4] dependence: the uncontrolled variable in free [T.sub.4] assays. Exp Clin Endocrinol 1994;102: 102-9.
(21.) Nelson JC, Nayak SS, Wilcox RB. Variable underestimates by serum free thyroxine ([T.sub.4]) immunoassays of free [T.sub.4] concentrations in simple solutions. J Clin Endocrinol Metab 1994;79: 1373-5.
(22.) Nelson JC, Weiss RM, Wilcox RB. Underestimates of serum free thyroxine ([FT.sub.4]) concentrations by free [T.sub.4] immunoassay. J Clin Endocrinol Metab 1994;79:76-9.
(23.) Nelson JC, Wilcox RB. Analytical performance of free and total thyroxine assays. Clin Chem 1996;42:146-54.
(24.) Wang R, Nelson JC, Weiss RM, Wilcox RB. Accuracy of free thyroxine measurements across natural ranges of thyroxine binding to serum proteins. Thyroid 2000;10:31-9.
(25.) Ellis M, Ekins RP. The radioimmunoassay of serum free triiodothyronine and thyroxine. In: Pasternak CA, ed. Radioimmunoassay in clinical chemistry. London: Heyden, 1975:299-306.
(26.) Giles AF. An improved method for the measurement of free thyroxine in serum dialysates. Clin Endocrinol (Oxf) 1982;16: 101-5.
(27.) Helenius T, Liewendahl K. Improved dialysis method for free thyroxin in serum compared with five commercial radioimmunoassays in nonthyroidal illness and subjects with abnormal con centrations of thyroxin-binding globulin. Clin Chem 1983;29: 816-22.
(28.) Larsen PR. Salicylate-induced increases in free triiodothyronine in human serum. Evidence of inhibition of triiodothyronine binding to thyroxine-binding globulin and thyroxine-binding prealbumin. J Clin Invest 1972;51:1125-34.
(29.) Tabachnick M, Hao Y-L, Korcek L. Effect of oleate, diphenylhydantoin and heparin on the binding of 1251-thyroxine to purified thyroxine-binding globulin. J Clin Endocrinol Metab 1972;36: 392-4.
(30.) Midgley JEM, Wilkins TA, Giles AF. Treatment with heparin and results for free thyroxin: an in vivo or an in vitro effect? Clin Chem 1982; 28:2441-3.
(31.) Bayer MF. Effects of heparin on serum free thyroxine linked to post-heparin lipolytic activity. Clin Endocrinol (Oxf) 1983;19: 591-6.
(32.) Stockigt JR, Lim CF, Barlow JW, Stevens V, Topliss DJ, Wynne KN. High concentrations of furoseamide inhibit serum binding of thyroxine. J Clin Endocrinol Metab 1984;59:62-6.
(33.) Lim CF, Bai Y, Topliss DJ, Barlow JW, Stockigt JR. Drug and fatty acid effects on serum thyroid hormone binding. J Clin Endocrinol Metab 1988;67:682-8.
(34.) Clark F, Horn DB. Assessment of thyroid function by the combined use of the serum protein bound iodine and resin uptake of 1311-triiodothyronine. J Clin Endocrinol Metab 1965;25:39-45.
(35.) Sutherland RL, Simpson-Morgan MW. The thyroxine-binding properties of serum proteins. A competitive binding technique employing Sephadex G-25. J Endocrinol 1975;65:319-22.
(36.) Nilsson SF, Peterson PA. Studies on thyroid-hormone-binding protein. I. The subunit structure of human thyroxine-binding globulin and its interaction with ligands. J Biol Chem 1975;250: 8543-53.
(37.) Korcek L, Tabachnick M. Thyroxine-protein interactions. Interaction of thyroxine and triiodothyronine with human TBG. J Biol Chem 1976;251:3558-62.
(38.) Snyder SM, Cavalieri RR, Goldfine D, Ingbar SH, Jorgensen EC. Binding of thyroid hormones and their analogues to thyroxinebinding globulin. J Biol Chem 1976;251:6489-94.
(39.) Carter JN, Eastman CJ, Corcoran JM, Lazarus L. Effect of severe, chronic illness on thyroid function. Lancet 1974;2:971-4.
(40.) Wood DG, Cyrus J, Samols E. Low [T.sub.4] and low [FT.sub.4]1 in seriously ill patients [Concise Communication]. J Nucl Med 1980;21:432-5.
(41.) Chopra IJ, Solomon DH, Gershon W, Hepner G, Morgenstein AA. Misleadingly low free thyroxine index and usefulness of reverse triiodothyronine measurement in non-thyroidal illnesses. Ann Intern Med 1979;90:905-12.
(42.) Chopra IJ, Van Herle AJ, Chua Teco GN, Nguyen AH. Serum free thyroxine in thyroidal and nonthyroidal illnesses: a comparison of measurements by radioimmunoassay, equilibrium dialysis and free thyroxine index. J Endocrinol Metab 1980;51:135-43.
(43.) Kurtz A, Dwyer K, Ekins RP. Serum free thyroxine in pregnancy [Letter]. Br Med J 1979;ii:550-1.
(44.) Whitworth AS, Midgley JEM, Wilkins TA. A comparison of free [T.sub.4] and the ratio of total [T.sub.4] to [T.sub.4]-binding globulin in serum through pregnancy. Clin Endocrinol (Oxf) 1982;17:307-13.
(45.) Ball R, Freedman DB, Holmes JC, Midgley JEM, Sheehan CP. Low-normal concentrations of free thyroxin in serum in late pregnancy: physiological fact not technical artifact. Clin Chem 1989;35:1891-6.
(46.) Midgley JEM. Methodological background of the Amersham Amerlex free thyroxine RIA. Nuklearmedizin 1982;21:174-83.
(47.) Symons RG, Vining RF. An evaluation of a fluorescence polarization immunoassay of thyroxine and thyroxine-uptake. Clin Chem 1985;31:1342-8.
(48.) Mendel CM. Preliminary evaluation of a fluorescence polarization immunoassay (Abbott TDx) for estimating serum free thyroxine concentrations in patients with critical nonthyroidal illness and low total thyroxine concentrations in serum [Technical Brief]. Clin Chem 1992;38:1916-7.
(49.) Hertl W, Odstrchel G. Kinetic and thermodynamic studies of antigen-antibody interactions in heterogeneous reaction phases. I. L-Thyroxine ([T.sub.4]) with specific antibody immobilised on controlled pore glass. Mol Immunol 1979;16:173-8.
(50.) Ashkar FS, Buehler RJ, Chan T, Howrani M. Radioimmunoassay of free thyroxine with pre-bound anti-[T.sub.4] microcapsules. J Nucl Med 1979;20:956-60.
(51.) Ekins RP, Filetti S, Kurtz AB, Dwyer K. A simple general method for the assay of free hormones (and drugs); its application to the measurement of serum free thyroxine levels and the bearing of the results on the "free thyroxine" concept. J Endocrinol 1980; 85:29P-30P.
(52.) Midgley JEM, Wilkins TA. Hypothyroidism in patients on fenclofenac. Lancet 1980;ii:704-5.
(53.) Wilkins TA, Midgley JEM, Stevens RAJ, Caughey I, Barron N. Assay performance and tracer properties for two analog-based assays of free triiodothyronine. Clin Chem 1986;32:465-9.
(54.) Midgley JEM, Wilkins TA. An improved method for the estimation of the relative binding constants of [T.sub.4] and its analogues with serum proteins. Clin Endocrinol (Oxf) 1982;17:523-8.
(55.) Amino N, Nishi K, Nakatani K, Mizuta H, Ichihara K, Tanizawa 0, Miyai K. Effect of albumin concentration of the assay of serum free thyroxine by equilibrium radioimmunoassay with labeled thyroxin analog (Amerlex free [T.sub.4]). Clin Chem 1983;29:321-5.
(56.) Bayer MF. Free thyroxine results are affected by albumin concentrations and non-thyroidal illness. Clin Chim Acta 1983;130: 391-6.
(57.) Stockigt JR, Stevens V, White EL, Barlow JW. "Unbound analog" radioimmunoassays for free thyroxine measure the albumin-bound hormone fraction. Clin Chem 1983;29:1408-10.
(58.) Midgley JEM, Winton MRJ, Wilkins TA. Relationship between effects of added albumin, initial free thyroxine value and endogenous serum-binding protein concentrations on Amerlex free thyroxine estimations. Clin Chim Acta 1987;167:67-79.
(59.) Ekins RP. Free hormones in blood: the concept and the measurement. J Clin Immunoassay 1984;7:163-80.
(60.) Ekins R. Validity of analog free thyroxin immunoassays [Opinions and Responses]. Clin Chem 1987;33:2137-52.
(61.) Midgley JEM, Wilkins TA. Effect of albumin concentration on equilibrium radioimmunoassay of serum free thyroxin with labeled thyroxin analog (Amerlex Free [T.sub.4]) [Letter]. Clin Chem 1983; 29:1861-3.
(62.) Wilkins TA, Midgley JEM. Analysis of Mass Action model for free thyroxin assays. Clin Chem 1986;32:566-8.
(63.) Midgley JEM. The free thyroid hormone hypothesis and measurement of free hormones [Letter and Response]. Clin Chem 1993;39:1342-4.
(64.) Christofides ND, Sheehan CP, Midgley JEM. One-step, labeled-antibody assay for measuring free thyroxin. I. Assay development and validation. Clin Chem 1992;38:11-8.
(65.) Christofides ND, Sheehan CP. Enhanced chemiluminescence labeled-antibody immunoassay (Amerlite-MAB) for free thyroxine: development and technical validation. Clin Chem 1995;41:17-23.
(66.) Midgley JEM, Sheehan CP, Christofides ND, Fry JE, Browning D, Mardell R. Concentrations of free thyroxin and albumin in serum in severe nonthyroidal illness: assay artifacts and physiological influences. Clin Chem 1990;36:765-71.
(67.) Seghers J, De Nayer P, Beckers C. Clinical evaluation of Amerlite's one-step immunoassay for determining free thyroxin [Technical Brief]. Clin Chem 1988;34:2160.
(68.) Reilly CP, Adamek KJ, Wellby ML. Critical assessment of the Amerlite free [T.sub.4] assay. Ann Clin Biochem 1989;26:517-21.
(69.) John R, Henley R, Shankland D. Concentrations of free thyroxine and free triiodothyronine in serum of patients with thyroxin- and triiodothyronine-binding autoantibodies. Clin Chem 1990;36: 470-3.
(70.) Sapin R, Gasser F, Schlienger JL, Chambron J. Analytical and clinical evaluation of a new one-step non-analogue radioimmunoassay for serum free thyroxine. Eur J Nucl Med 1989;17:111-5.
(71.) Sapin R, Gasser F, Schlienger JL. Familial dysalbuminemic hyperthyroxinemia and thyroid hormone autoantibodies: interterence in current free thyroid hormone assays. Horm Res 1996; 45:139-41.
(72.) Giacomini KM, Swezey SE, Giacomini JC, Blaschke TF. Administration of heparin causes in vitro release of nonesterified fatty acids in human plasma. Life Sci 1980;27:771-80.
(73.) Riemersma RA, Russell DC, Oliver MF. Heparin-induced lipolysis; an exaggerated risk. Lancet 1981;ii:471.
(74.) Mendel CM, Frost PH, Cavalieri RR. Effect of free fatty acids on the concentration of free thyroxine in human serum: the role of albumin. J Clin Endocrinol Metab 1986;63:1394-9.
(75.) Mendel CM, Frost PH, Cavalieri RR. In vitro lipolysis during equilibrium dialysis may cause overestimation of free thyroxine levels [Abstract]. Clin Res 1986;34:429A.
(76.) Mendel CM, Frost PH, Kunitake ST, Cavalieri RR. Mechanism of the heparin-induced increase in the concentration of free thyroxine in plasma. J Clin Endocrinol Metab 1987;65:1259-64.
(77.) Juame JC, Mendel CM, Frost PH, Greenspan FS, Laughton CW. Extremely low doses of heparin release lipase activity into the plasma and can thereby cause artifactual elevation in the serum free thyroxine concentrations as measured by equilibrium dialysis. Thyroid 1996;6:79-83.
(78.) Stevenson HP, Archbold GPR, Johnston P, Young IS, Sheridan B. Misleading serum free thyroxine results during low molecular weight heparin treatment. Clin Chem 1998;44:1002-7.
(79.) Hershman JM, Krugman LG, Kapple JD, Reed AW, Azukizawa M, Shinaberger JH. Thyroid function in patients undergoing maintenance hemodialysis; unexplained low serum thyroxine concentration. Metabolism 1978;27:755-9.
(80.) Braverman LE, Abreau CM, Brock P, Kleinmann R, Fournier L, Odstrchel G, Shoemaker HJ. Measurement of serum free thyroxine by RIA in various clinical states. J Nucl Med 1980;21:233-9.
(81.) Bayer MF, McDougall IR. Radioimmunoassay of free thyroxine in serum: comparison with clinical findings and results of conventional thyroid function tests. Clin Chem 1980;26:1186-92.
(82.) Kaptein EM, Mclntyre SS, Weiner JM, Spencer CA, Nicoloff JT. Free thyroxine estimates in nonthyroidal illness: comparison of eight methods. J Clin Endocrinol Metab 1981;52:1073-7.
(83.) Kaptein EM, Grieb DA, Spencer CA, Wheeler WS, Nicoloff JT. Thyroxine metabolism in the low thyroxine state of critical nonthyroidal illness. J Clin Endocrinol Metab 1981;53:764-71.
(84.) Slag MF, Morley JE, Elson MK, Labrosse KR, Crowson TW, Nuttall FQ, Shafer RB. Free thyroxine level in critically ill patients: a comparison of currently available assays. JAMA 1981;246: 2702-6.
(85.) Melmed S, Geola FL, Reed AW, Pekary AE, Hershman JM. A comparison of methods for assessing thyroid function in nonthyroidal illness. J Clin Endocrinol Metab 1982;54:300-6.
(86.) Wehmann RE, Gregerman RI, Burns WH, Sarai R, Santos GW. Suppression of thyrotropin in the low thyroxine state of severe nonthyroidal illness. N Engl J Med 1985;312:546-52.
(87.) Brent GA, Hershman JM, Braunstein GD. Patients with severe nonthyroidal illness and serum thyrotropin concentrations in the hypothyroid range. Am J Med 1986;81:463-6.
(88.) Bayer MF, Macoviak JA, McDougall IR. Diagnostic performance of sensitive measurements of serum thyrotropin during severe nonthyroidal illness: their role in the diagnosis of hyperthyroidism. Clin Chem 1987;33:2178-84.
(89.) Eggertsen R, Petersen K, Lundberg PA, Nystrom E, Lindstedt G. Screening for thyroid disease in a primary care unit with a thyroid stimulating hormone assay with a low detection limit. Br Med J 1988;297:1586-92.
(90.) Chopra IJ, Chua Teco GN, Nguyen AH, Solomon DH. In search of an inhibitor of thyroid hormone binding to serum proteins in nonthyroidal illness. J Clin Endocrinol Metab 1979;49:64-9.
(91.) Woeber KA, Maddux BA. Thyroid hormone binding in nonthyroidal illness. Metabolism 1981;30:412-6.
(92.) Oppenheimer JH, Schwartz HL, Mariash CN, Kaiser FE. Evidence for a factor in the sera of patients with nonthyroidal disease which inhibits iodothyronine binding by solid matrices, serum proteins and rat hepatocytes. J Clin Endocrinol Metab 1982;54: 757-66.
(93.) de Groot U. Dangerous dogmas in medicine: the nonthyroidal illness syndrome [Review]. J Endocrinol Metab 1999;84:151-64.
(94.) Csako G, Zweig MH, Glickman J, Ruddel M, Kestner J. Direct and indirect techniques for free thyroxin compared in patients with nonthyroidal illness. II. Effect of albumin, prealbumin and thyroxin-binding globulin. Clin Chem 1989;35:1655-62.
(95.) Vagenakis AG, Portnay GI, O'Brian JT, Rudolph M, Arky RA, Ingbar SH, Braverman LE. Effect of starvation on the production and metabolism of thyroxine and triiodothyronine in euthyroid obese patients, J Clin Endocrinol Metab 1977;45:1305-9.
(96.) Balsam A, Ingbar SH. The influence of fasting, diabetes and several pharmacological agents on the pathways of thyroxine metabolism in rat liver. J Clin Invest 1978;62:415-24.
(97.) Stockigt JR. Guidelines for diagnosis and monitoring of thyroid disease: nonthyroidal illness. Clin Chem 1996;42:188-92.
(98.) Lamb EJ, Martin J. Thyroid function tests: often justified in the acutely ill. Ann Clin Biochem 2000;37:158-64.
(99.) Schatz DL, Sheppard RH, Steiner G, Chandarlapaty CS, de Veber GA. Influence of heparin on serum free thyroxine. J Clin Endocrinol Metab 1969;29:1015-22.
(100.) Wang YS, Hershman JM, Smith V, Pekary AE. Effect of heparin on free thyroxin as measured by equilibrium dialysis and ultrafiltration. Clin Chem 1986;32:700.
(101.) Jain R, Uy HL. Increase in serum free thyroxine levels related to intravenous heparin treatment. Ann intern Med 1996;12:74-5.
(102.) Spector AA. Fatty acid binding to plasma albumin [Review]. J Lipid Res 1975;16:165-79.
 Nonstandard abbreviafions: [T.sub.4], thyroxine; [T.sub.3], triiodothyronine; [FT.sub.4], Tree [T.sub.4]; TBG, thyroxine-binding globulin; FTI, Iree thyroxine index; NEFA, nonesterified Tatty acid; NTI, nonthyroidal illness; BP, binding potential; and TSH, thyrotropin.
JOHN E.M. MIDGLEY
North Lakes Clinical, 6 High Wheatley, Ilkley, West Yorkshire LS29 SRX, United Kingdom. E-mail firstname.lastname@example.org.
|Printer friendly Cite/link Email Feedback|
|Author:||Midgley, John E.M.|
|Date:||Aug 1, 2001|
|Previous Article:||Analysis issues for gene expression array data.|
|Next Article:||Neonatal screening for galactosemia by quantitative analysis of hexose monophosphates using tandem mass spectrometry: a retrospective study.|