Immunohistochemistry for the Hematopathology Laboratory

Antigens

At the heart of IHC is the antigen-antibody reaction; therefore, it is crucial that the antigenic epitopes recognized by the cognate diagnostic antibody maintain their reactive conformation. The specific antigenic epitopes present on any given protein or carbohydrate moiety are subject to enzymatic degradation that begins immediately after biopsy or resection and to further conformational changes resulting from fixation. To ensure preservation of the antigen of interest, rapid tissue fixation is important. Some antigenic epitopes, such as those on keratin proteins and other structural proteins of the cell, are relatively resistant to degradation; other antigens, such as phosphoepitopes on signaling proteins, undergo rapid degradation within minutes to hours.

Although prompt tissue fixation is essential to preserve antigenicity, the specific fixative and the fixation process itself can interfere with antigenicity by causing conformational changes in antigenic molecules or by actually chemically modifying the antigenic epitopes. Traditionally, tissues have been fixed in neutral buffered formalin (pH 7.0) because it is inexpensive, has sterilizing properties, and preserves morphologic features well. The exact chemical reactions that occur in tissues are not well understood, but it is generally assumed that formalin's ability to cross-link aldehyde groups in proteins is responsible for its fixative properties. This mode of action is potentially deleterious to antigenic structure, and although some antigenic epitopes may not be affected significantly by formaldehyde cross-linking, these chemical modifications clearly have an adverse effect on many antigens. Because formalin penetrates tissues slowly and the chemical reactions are complex, the number of modifications that take place is time dependent. In practice, this means that antigens fall into three basic categories: formalin-resistant epitopes, highly formalin-sensitive epitopes, and epitopes with a time-dependent sensitivity to formalin fixation. Although there have been attempts to generate antibodies specifically to formalin-resistant epi­topes, most of the antibodies found to react with formalin-resistant epitopes have been identified through large-scale screenings of available antibody preparations.

Over the years, there has been great interest in identifying methods to overcome or reverse the deleterious effects of formalin fixation. The earliest attempts to retrieve antigenicity used proteolytic enzymes, which presumably act by breaking formaldehyde-induced methylene cross-links in the antigenic molecules, thereby relaxing some of the conformational constraints on the protein epitopes. Such proteolytic methods continue to be used in many IHC laboratories and are particularly useful for recovering the reactivity of the cytokeratins. Nonetheless, proteolytic methods are difficult to control, and careful attention is needed to optimize their retrieval effect and avoid tissue destruction.

Despite some successes with proteolytic methods, the major breakthrough that brought IHC into widespread use was the development of heat-induced epitope retrieval (HIER) procedures. This technique involves heating fixed tissue sections in buffered solutions at or above 100° C for several minutes to more than 30 minutes. HIER methods vary in terms of the recommended buffer solutions and the mode of heating, but the basic formula of applying wet heat over a period of time is universal. The exact mechanism by which HIER reverses the loss of antigenicity in formalin-fixed tissue is unknown. However, hydrolytic cleavage of formaldehyde-related chemical groups and cross-links, the unfolding of inner epitopes, and the extraction of calcium ions from coordination complexes with proteins are among the hypothesized mechanisms.

The advent of HIER methods revolutionized IHC and greatly expanded the number of antibodies that react in formalin-fixed, paraffin-embedded tissue sections. HIER has also improved the sensitivity of antibodies directed to formalin-resistant epitopes and has enabled the routine assessment of a wide spectrum of antigens in epoxy resin–embedded bone marrow sections. Appropriate antigen retrieval can minimize many of the problems related to preanalytic factors, reducing differences in immunostaining that result from the variations in fixation time in the clinical laboratory.

The major disadvantage of HIER is that the high heat can cause considerable tissue damage, particularly when the tissue is underfixed or has a high collagen content, the antigen-retrieval time is prolonged, and the buffers contain ethylenediaminetetraacetic acid (EDTA) or have a high pH. Tissue damage can be minimized by ensuring that tissues are optimally fixed, reducing the antigen-retrieval time, or changing the retrieval buffer. Despite this potential problem, the ability to detect otherwise non-detectable antigens far outweighs the potential for tissue damage on occasional tissue sections.

Primary Antibodies

There are two major categories of primary antibodies used in diagnostic pathology: monoclonal antibodies and polyclonal antibodies. Polyclonal antibodies are generated by injecting an animal (most commonly a rabbit or goat) with the antigenic preparation of interest and harvesting the animal's serum once an immune response is detected. The serum is subjected to antibody purification and sometimes to differential adsorptions to eliminate unwanted reactivity, but it always comprises a spectrum of antibody molecules originating from multiple unrelated antibody-producing cells (hence the term polyclonal ). The specificity of a polyclonal antibody preparation is highly dependent on the purity of the initial antigenic preparation and how extensively adsorbed it is. Obtaining highly specific preparations is difficult, and background problems can be troublesome, especially when applied to IHC. Further, because the antibody response is variable over time and from one individual animal to another, complete standardization of antibody composition is not possible. Although developments in recombinant DNA and protein synthesis technology have greatly improved the specificity of polyclonal antibodies by providing tools to generate highly purified protein immunogens or even specific immunogenic peptides, polyclonal antibodies may still contain unwanted specificities.

Monoclonal antibodies, in contrast, are the product of a single immortalized antibody-producing cell, thus avoiding most problems related to antibody heterogeneity and specificity inherent in polyclonal antibody preparations. The hybridoma technology pioneered by Kohler and Milstein in the 1970s allows the immortalization of a single antibody-producing mouse plasma cell by fusing it with a mouse plasmacytoma cell line. Individual hybrid mouse cells can be clonally expanded in tissue culture or in mice as tumors, providing a continuous source of antibody of known composition and reactivity. Because of their high quality and specificity, monoclonal antibodies were rapidly developed as diagnostic reagents in hematopathology, as well as for other clinical applications that require standardized reagents. The specificity advantage of the monoclonal antibody, however, can also be a disadvantage when applied to denatured proteins in tissue sections. Because a polyclonal antibody preparation generally contains a mixture of antibodies reacting to multiple epitopes, it does not matter if some of the epitopes are rendered inactive by the fixation process, as long as one epitope remains in its reactive conformation. However, if the single epitope recognized by a monoclonal antibody is affected by the fixation process, the antibody cannot be used for IHC. A second disadvantage of the mouse monoclonal antibodies is that they generally have weaker affinity constants than do comparable polyclonal rabbit antibody preparations. This led to the development of rabbit plasmacytoma cell lines that could be used as fusion partners to generate high-affinity rabbit monoclonal antibodies. These high-affinity rabbit monoclonal antibodies have improved the detection of some antigens, such as cyclin D1, and permitted detection of others that were heretofore unavailable with murine antibodies, such as CD103. Rabbit monoclonal antibodies are now available for many targets of hematopathologic interest, including CD3, CD5, CD8, CD23, CD56, CD79a, CD103, cyclin D1, and Ki-67.

Regardless of which type of antibody is chosen for an immunohistochemical procedure, careful control over the development and use of the antibody must be maintained. Although antibody specificity is best demonstrated by immunoblotting or immunoprecipitation, this type of biochemical analysis is required only during the initial development of the antibody. However, before placing any antibody into clinical use, extensive validation of its efficacy and staining characteristics on tissue sections in the individual laboratory is necessary. This should include extensive testing of normal and tumor tissues to assess the specificity and sensitivity of tissue staining. The use of tissue microarrays can be extremely helpful during this stage. Once the antibody has been validated and placed in service, the continued use of both negative and positive controls is mandatory with each test sample. Negative controls are best demonstrated by omitting the primary antibody or by substituting the specific primary antibody with an isotype-matched control antibody or immunoglobulin (Ig) preparation. Positive controls should include tissues that are known to contain the antigen of interest.

Detection Systems

Detection systems comprise an enzyme, a chromogenic substrate, and a link or bridge reagent that brings the enzyme into proximity with the primary labeling antibody. The choice of a detection system is of great importance, and each method has its own advantages and disadvantages ( Table 4-1 ). Factors influencing the selection of a detection method are related to the type of tissue, the cellular target, its abundance and localization, and laboratory-specific issues (e.g., complexity, time requirements, reagent costs). The most widely used detection systems today are the biotin-based systems—of which the avidin-biotin immunoperoxidase complex (ABC) system developed by Hsu and coworkers may be considered a prototype—and the more recently developed polymer-based systems. In the ABC system, a tissue-bound primary unlabeled antibody is reacted with a secondary biotin-conjugated link antibody, and detection is carried out through preformed avidin-biotinylated enzyme (peroxidase) complexes. The peroxidase enzyme in the complex then reacts with a chromogen (e.g., 3,3′-diaminobenzidine [DAB] or 3-amino-9-ethylcarbazole [AEC]) to produce a colored reaction product that is discretely localized to the targeted antigen. More recently, polymer-based detection systems have been developed that do not depend on avidin-biotin links, thereby avoiding the possibility of high backgrounds in tissues rich in endogenous biotin. Like in the biotin-based systems, an unlabeled primary antibody is used first, followed in this case by a modified polymer (e.g., dextran) that is linked to a large number of secondary link antibodies and enzyme (peroxidase) molecules. Thus, one reagent contains both a species-specific secondary anti-immunoglobulin linking antibody and the chromogen developing enzyme. Newer detection systems have also been developed to increase the sensitivity for detecting antigens expressed at very low levels or to improve the detection of low-affinity primary antibodies. These systems involve a tyramide-based signal amplification method known as the catalyzed reporter deposition (CARD) or catalyzed system amplification (CSA) method.

Interpretive Problems

It is necessary to distinguish specific from non-specific signals when interpreting IHC. There are many sources of false-positive results, including endogenous biotin or peroxidase, inappropriately high antibody concentrations, poor technique (e.g., excessive antigen retrieval, drying artifacts, prolonged detection), or interpretive errors such as mistaking endogenous pigment for the chromogenic reaction product. Endogenous biotin reactivity can be a serious problem because of its variable occurrence in tumors. This biotin positivity is often amplified by retrieval techniques and displays a granular pattern that can be difficult to distinguish from other granular cytoplasmic staining. Failure to block biotin can lead to problems with interpretation and the reporting of false-positive results. Use of one of the newer polymer-based detection systems that avoids the use of a biotin-avidin link can eliminate this problem. False-negative results also have myriad reasons, the most frequent of which are inadequate antigen retrieval, suboptimally fixed tissue, inappropriate primary antibody, or other technical staining issues.

It cannot be overemphasized that the accurate interpretation of IHC stains requires knowledge of the laboratory's methods, the antibodies used, and the expected staining pattern for each antibody. Different antibody preparations to the same antigen may show various patterns and intensities of non-specific or even specific staining. For instance, the traditional polyclonal carcinoembryonic antigen (CEA) antibodies cross-react with other CEA-like proteins such as CEACAM6 and stain granulocytes, whereas specific monoclonal CEA antibodies do not. Monoclonal antibodies targeting different epitopes of the TREG-associated marker FOXP3 have been shown to stain different subpopulations of cells in comparative studies in paraffin sections. As another example, the anti–Ki-67 monoclonal antibody MIB-1 has been reported to stain the cell membrane of some tumor types, whereas other monoclonal antibodies to the same antigen do not show this type of aberrant staining. Knowledge of the subcellular staining location of the targeted antigen is crucial. There are a number of expected locations for antibody signals, including nuclear, cytoplasmic, membranous, Golgi, and extracellular as well as combinations of these stereotypical patterns ( Fig. 4-1 ). An unexpected staining localization should immediately raise a red flag and should not be considered positive in any situation. For example, in a recent assessment of synaptophysin antibodies by the NordiQC organization, one of several monoclonal antibody preparations was found to produce an unusual dotlike staining reaction in tissues that were known to be negative for synaptophysin. This artifactual staining pattern was thought to be the result of a cross-reaction with a Golgi-associated protein—an artifact that was previously associated with other monoclonal antibodies prepared from mouse ascites, as was the case for this particular antibody. It is also critical that the interpreter be able to distinguish non-specific background staining or pigment deposits from true staining resulting from the presence of the antigen. It is the ultimate responsibility of the hematopathologist to be familiar with the methods and specific antibodies used by the laboratory, as well as the expected staining patterns of the targeted antigens when using these results to provide diagnoses.