Point: Legitimate and Illegitimate Tests of Free-Analyte Assay Function

In the history of clinical immunoassay, even up to the present day, no group of tests has been subjected to more scrutiny and controversy than analog-type free-analyte assays (1)(2)(3)(4). Unfortunately, inappropriate design of many experiments has led users to an incorrect perspective of how analog assays work. An analysis of the important series of studies by Nelson and coworkers (5)(6)(7)(8)(9)(10) shows that most experiments were conducted outside the valid working limits of analog-type or 2-step free-analyte assays. Here we explore the legitimacy of this viewpoint. In gold-standard methods of equilibrium dialysis and ultrafiltration, free analyte is first separated from that bound to serum proteins. Only then is the free analyte measured by what essentially is a high-sensitivity total analyte assay. For these methods, testing of assay parameters such as recovery of added analyte, linearity of response, and cross-reactivity are legitimate, applying equally to total analyte measurements. When the antibody probe is in direct contact with serum, however, the situation is radically different. For bound thyroxine (T4), serum contains a low-concentration, high-affinity binding protein (thyroxine-binding globulin) and high-concentration, low-affinity proteins (transthyretin and albumin). Free and total analyte are nonlinearly related (except with zero-thyroxine-binding globulin serum) (11). Recovery and linearity experiments in such conditions are thus not feasible, becoming increasingly less accurate at higher thyroxine-binding globulin concentrations (11). Analog-type and equilibrium dialysis/ultrafiltration methods must be calibrated differently. With gold-standard assays, direct calibration is done gravimetrically with known quantities of analyte, a method that is not possible with analog assays. Analog-type methods are calibrated by secondarily applying the values originally obtained from gold-standard methods to the same sera. The use of this calibration method is unavoidable, because much more analyte is bound to the antibody probe in the presence than the absence of the bound-analyte reservoirs of a given serum (12). The calibration of analog-type assays is analogous to water pressure measurement in a pipe from a reservoir, as illustrated in Fig. 1 . The figure represents a reservoir that is connected to a thin pipe B of comparatively negligible volume (<0.1% of the total). The reservoir size can vary greatly, with the flow through the pipe depending only on the pressure provided by height H of the water. The reservoir equates to analyte bound to variable amounts of serum proteins, and pipe B to free analyte entering the tissues. The higher the reservoir level, the greater is the flow and vice versa, equating to hyper or hypo analyte status in serum. Analogy of a serum system after blood separation, when free analyte is assayed by either equilibrium dialysis or analog-type methods. (A), A is a manometer attached to the vertical pipe B (acting as a sampling device for the system—equivalent to assay conditions and antibody sampling). Measurement of water level with reservoir in contact with manometer (equivalent to the working of analog-type or 2-step free-T4 (FT4) assays). (B), Representation of equilibrium dialysis/ultrafiltration. Measurement of water when pipe B is first detached from reservoir before measurement by manometer (equivalent to measuring FT4 by equilibrium dialysis or ultrafiltration). If a thin manometer A is attached to the system, water will flow up the manometer to a height that is level with the reservoir. Owing to the negligible volume of both pipe B and manometer A, the reservoir’s water level scarcely changes (Fig. 1A ). Thus, for various levels, calibrations can be marked on the manometer. This mirrors a free-analyte analog assay, the manometer being equivalent to added antibody probe and calibration marks equivalent to measured free analyte concentration. In an alternative situation, the pipe B is disconnected from the reservoir before attachment to manometer A (Fig. 1B ). In this case, because there is no reservoir replenishing water in B as it enters the manometer, the final equilibrium position in the manometer rests much lower. This is equivalent to the performance of analog assays when only pure solutions of analyte are added (i.e., gross underestimation of the water pressure, and equivalently, the free analyte). The use of such pure solutions to analyze analog assays is invalid (5)(6)(7)(8)(9)(10). If the reservoir level is unduly affected (lowered) by manometer sampling, invalidity also occurs, owing to overly large sequestration of fluid. Likewise, challenging overdiluted serum with a fixed amount of antibody results in undue disturbance of the natural equilibrium between free and bound analyte (12). The pipe B, disconnected from the reservoir before measurement of fluid by the manometer, is equivalent to equilibrium dialysis or ultrafiltration, in which the free analyte is separated from serum before assaying the dialyzed analyte. Both sets of calibrations are internally consistent, but only one applies when the reservoir plays its part in determining the fluid level in the manometer. The 2 methods are calibrated completely differently, although each calibration in one method is proportional to its partner in the other. This simple analogy shows nonmathematically why the studies of Nelson and coworkers (5)(6)(7)(8)(9)(10) cannot address how analog-type assays (or indeed any 2-step free analyte assays) work. Such assays have a limited “window of validity” wherein meaningful experiments negligibly disturb the natural equilibrium between serum protein-bound and free analyte. If this criterion is not obeyed in cases in which pure analyte solutions are used or in intermediate situations, the results are not legitimate. Again, however, there are different criteria to consider. With equilibrium dialysis and ultrafiltration, the dialysis buffer dilutes the test serum during separation of bound and free analyte. Thus, in this case, dilution is the important factor in controlling constancy of free-analyte measurement. For analog-type and 2-step assays, both dilution and the simultaneous sequestration of analyte from its bound state by the antibody probe are equally important. Nevertheless, both methodologies are constrained by the limits of dilution permissible without unduly affecting the concentration of bound analyte. Gold standard methods are more robust to dilution, owing to the absence of the additional sequestration effect of the antibody probe. With analog assays, dilutions >6-fold may exceed permitted validity limits, especially in sera with low binding capacities. For all versions of free analyte assays, it is legitimate to examine the effects of spiking serum with substances capable of displacing bound T4. However, the results will vary depending on the serum used. In either gold-standard or analog-type assays, addition of nonesterified fatty acids to serum should change (increase) measured free analyte concentrations only after they have saturated all albumin-binding sites. The point and the rate at which this occurs depend very much on the concentration of the appropriate binding proteins in a given serum and cannot be easily predicted. Likewise, addition ofextra albumin will steadily reduce the measured free-analyte concentration, but the extent of this decrease also depends on the individual serum used and, especially, the properties of added albumin (13). Finally, well-designed analog-type free-analyte immunoassays no longer suffer the protein effects that affected initial designs (14)(15). Attempts to show otherwise (5)(6)(7)(8)(9)(10) are essentially illegitimate unless the fundamental criterion of validity described above is obeyed. Such experiments are meaningless and have nothing useful to say about analog assay performance. Assays work legitimately only over a limited range of permitted concentrations and properties of ingredients, and analytic experiments must be conducted with respect to this fact.