The Evolution of Immunoassay as Seen Through the Journal Clinical Chemistry

A brief historical review shows that the first reference in the Journal to an immunologically based method was in 1962 [in a method for lipoproteins using an immunoprecipitation technique (1)]. In 1962 and 1963, there were two references to enzyme-mediated methods (again the first in the Journal) for glucose and oxalate! In 1965 came the first reference to a competitive protein binding method for thyroxine (2), which was also automated (!). In contrast, in 1966 there was a reference to the effect of ultrasound on isoenzymes! In 1968 came the second reference to an immunologically based assay, this time for growth hormone (3), as well as, in contrast, the use of nuclear magnetic resonance for studying low molecular weight constituents! I mention these other landmark papers because enzyme-mediated and protein-binding assays reflect other examples of biorecognition systems that have also matured with great success (particularly in the case of enzymes) as well as other analytical concepts that in the 1960s may have been considered ahead of their time, in relation to what has been achieved in laboratory medicine. In 1969, papers on immunoassay began to blossom, with more publications on the assessment of thyroid status, the issues of sensitivity in relation to quantification of low concentration proteins in cerebrospinal fluid, and reagent quality control. These early papers should be placed chronologically in the context of the publication on the radioimmunoassay of insulin by Yalow and Berson (4) in 1959 and the measurement of thyroxine by Ekins about that time. Thus, in the first 15 years of the Journal, there were approximately 10 papers on immunologically based assays or their utilization; in 1997 there were in excess of 100 papers in which immunoassay methods were utilized (>35% of the total number of papers published). Looking at the Editor’s perspective on the 30 most cited papers in the Journal (Clin Chem 1998;44:698–9), I find 4 that involve immunoassay; they carry important messages. The abstracts of the first three are shown as Figs. 1–3. The contribution by Rodbard (5) on data handling provides a detailed treatment of an aspect of immunoassay that lives in the shadow of the huge developments in basic knowledge of the antigen–antibody reaction and reagent technology (Fig. 1); the discussion has been continued over many years by Rodbard and colleagues as well as other contributors. In the field of reagent technology, the paper by Del Villano et al. (6) was one of the first in which an antibody was used to characterize a marker, in this instance the tumor marker CA 19-9 (Fig. 2); antibodies have been used to characterize specific antigen moieties on many occasions since, including prostate-specific antigen and nicked human chorionic gonadotropin. The paper by Nussbaum et al. (7) illustrated the clinical benefits of using two monoclonal antibodies against unique epitopes to quantify the whole molecule of parathyroid hormone (Fig. 3). This description by Nussbaum et al. (Clinical Chemistry 1987;33:1364–7) of a highly sensitive two-site IRMA for human parathyroid hormone had been cited more than 350 times by the end of 1995. The paper by Del Villano et al., describing the use of a monoclonal antibody to characterize the tumor marker CA 19-9, Clinical Chemistry 1983;29:549–52. By the end of 1995, this paper had been cited more than 300 times. Rodbard’s review of data handling in immunoassays (Clinical Chemistry 1974;20:1255–70) had been cited more than 800 times by the end of 1995. The foundation of any immunoassay must be its antibody and its complementary antigen. Knowledge of the chemistry of the antigen–antibody reaction, the contribution of different binding forces, the spatial orientation of critical elements of the molecular structure, and the ultimate definition of definitive epitope and paratope regions has been gleaned from fundamental techniques such as x-ray diffraction, nuclear magnetic resonance, elipsometry, and mass spectrometry (8), together with the experience gained from the clinical application of assays and their subsequent refinement. In this latter respect, the Journal has over the years carried a fascinating dialogue on the measurement of digoxin, which through a mixture of comparative studies (initially with chromatographic methods) and clinical experience led to the recognition of digoxin-like molecules. Careful selection of antibodies, development of more-specific immunoassay techniques for haptens, and more recently, the use of antiidiotypic antibodies have resulted in the demonstration of a more-specific assay (9). Knowledge of the detailed molecular structure of the antibody has led to developments in protein engineering aimed at modification of the binding characteristics. This is primarily achieved through the construction of a single-chain antibody; expression of this protein in a suitable vector enables the use of techniques such as site-directed mutagenesis to generate novel antibodies. An alternative approach to developing a biorecognition molecule is the construction of antibody mimics. An antigenic binding pocket is produced in a polymeric structure using chemical synthesis technology, obviating the need for any biological vector (10). The nature of the antigen has also played a crucial role in immunoassay performance, both in its role as an immunogen, as a labeled reactant in a competitive assay, and in its role as a calibrator. The design of an immunogen, exemplified by the use of a glycated peptide to generate monoclonal antibodies against HbAIC, has demonstrated the evolution of skills in reagent technology with the advent of synthetic immunogens (11). Perhaps the advent of antibody mimics heralds the conquering of the “black art” of antibody production, putting it firmly into the arena of chemical synthesis. The development of Emit (enzyme-multiplied immunoassay) announced a new dimension to immunoassay, exploration of the modulation of properties of the label attached to the antigen when an antibody molecule was bound. Other homogeneous technologies followed, including fluorescence polarization, substrate labeled fluorescence, fluorescence excitation transfer, cloned enzyme donor immunoassay [(CEDIA), see review by Gosling (12)], and luminescent oxygen channeling (LOCI) (13) to name but a few. Importantly, these technologies have also proven to be the critical enabling step in the development of aspects of modern clinical practice, perhaps the best example being therapeutic drug monitoring. The antigen is also a critical feature of method calibration and, thus, method bias. This is an area of practical science where the Journal has an unrivalled record of reporting. However, there are many instances where unacceptable method bias has gone unchallenged—the idea of a cross-calibration comparison not having been entertained. This notwithstanding, many early calibrator materials had method-specific matrix effects making such comparison unworkable; was this art or accident? Thus in 1976, Reimer and Maddison (14) discussed the problems of standardization of immunoglobulin assays; the topic arose again in 1978, and culminated in the report by Whicher et al. (15) on the preparation of an international reference preparation for several serum proteins. Several useful reviews in the Journal provide the details on developments of new labels and the objectives that have been achieved (12)(16). There has been a dramatic reduction in the lowest detectable concentration achievable, leading to the addition to the vocabulary of the yoctomole and zeptomole at a time when few people are even familiar with the attomole; several of the labels have now been used to achieve detection limits in the 10−17–10−19 mol/L region in routine assays, with the more popular examples including enzyme amplification, time-resolved fluorescence, chemiluminescence, enzyme-mediated luminescence, and electrochemiluminescence. One of the more unusual labels has been the use of DNA-labeled antibodies with the amplification properties of the polymerase chain reaction (17). The application of nonisotopic labels has facilitated the development of robust automation and assay reagents with a longer shelf, as well as the fabrication of assays into point-of-care devices and improvements in the clinical effectiveness of assays (more of that later). The automation of heterogeneous immunoassays has been a challenge to both the chemist and the engineer. Although the adaptation of the centrifugal analyzer was one of the first published attempts at automation, it was probably the work of Greenwood et al. (18) with the magnetic particle solid phase that was seminal to many automated systems. Early attempts used continuous flow technology [including the Southmead system (19)], but discrete analyzers with the capability for washing and capturing the magnetic solid phase have evolved in more recent years. The first point-of-care device was reported in 1979 (!) with an immunocapture migration assay for insulin (20), but many other innovative devices have followed—too many to quote in a short editorial. The devices described have been predominantly of three types: (a) flow through, (b) complex fabricated microcell, and (c) immunosensors. Interestingly, there are probably fewer publications on the concepts of the first two categories (21) but many commercialized examples, whereas there is a wealth of published work on the principles underlying immunosensor systems, but few commercialized examples (22). On the other hand, the immunosensor technology that has been commercialized has provided a valuable analytical tool for studying the real-time kinetics of the antigen–antibody reaction. Many of the excellent ideas and innovations in immunoassay often stumble, albeit temporarily, in their routine application, and the pages of the Journal are testimony to this phenomenon—interferences. No assay can be considered robust until it has been exposed to many hundreds of samples from patients with different diseases. There are several well-known interferents—including influences on the label, e.g., bilirubin, ascorbate, and endogenous substrates; antigen-like molecules, e.g., digoxin-like molecules, analyte fragments, and isoforms; and protein binding, e.g., paraproteins, autoantibodies, heterophilic antibodies, and carrier proteins. The intermethod comparison, although increasingly frowned on by editors, can provide valuable insight into the “fragility” of an assay; anyone wishing to research a single method in detail will find invaluable information in the pages of Clinical Chemistry often not presented anywhere else. The evolution of immunoassay has brought considerable benefits in terms of analytical, operational, and clinical outcomes. Thus, developments in digoxin methodology from liquid chromatography, through polyclonal and monoclonal antibody-based assays, to a pseudoimmunometric approach with a homogeneous format have facilitated rapid accurate quantification at the point of care. The story is very similar in the case of cortisol. In the case of markers of cardiac muscle damage, we have seen improvements in enzyme measurement, immunoinhibition assays for the creatine kinase MB isoenzyme, through to mass measurements. This inevitably led to the discovery of more-specific markers, with the advent of automated and point-of-care assays improving diagnostic accuracy, enabling improved patient triage, and helping to reduce lengths of stay in costly clinical facilities (23). Other examples include the evolution from phosphatase activity to free prostate-specific antigen, protein-bound iodine to thyrotropin, and urine glucose to glycohemoglobin. These case studies perhaps represent the true achievement of developments in immunoassay technology. So where does the future lie? In a 1976 review of enzyme immunoassay (24), G.B. Wisdom (Fig. 4) correctly predicted that improvements were certain. No doubt we will continue to see improvements in methods for established analytes in tandem with, in many cases, an enhanced appreciation of the clinical effectiveness of the diagnostic tests. In parallel with the advent of new tests, there will be pressure for more rapid delivery of results and thus, with it, the need to enhance the reaction rates to reduce assay times. Faster reaction tests require faster collisions—possibly a role for ultrasound—and, therefore, shorter diffusion distances—a role for miniaturization, most definitely [(25) and the Oak Ridge Conference Proceedings in last month’s issue of the Journal, e.g., References (26) and (27)]. Wisdom’s review predicting the improvements in enzyme immunoassays, Clinical Chemistry 1976;22:1243–55. This review had been cited more than 300 times by the end of 1995. This has been a personal perspective of immunoassay developments seen through the medium of the journal Clinical Chemistry. The challenges illustrated in the evolution still remain in scientific terms, with perhaps greater emphasis today on striving for the recognition of value for money and proper investment in diagnostic tools to realize the clinical and operational benefits that have been demonstrated in the Journal. This challenge is exemplified in recent papers (28)(29), which showed that point-of-care testing in the emergency department reduced the time taken for the doctor to obtain results but did not influence the overall length of stay in the emergency room!