The Total Test Approach to Standardization of Immunohistochemistry

Histopathology continues to serve as the mainstay for diagnostic and therapeutic decisions in almost all forms of neoplasia. Yet, as previously described,1 surgical pathologists in >routine practice have problems recognizing many of the more subtle criteria, and even experts are far from infallible.2 It has been written that “All pathologists would probably agree that the interpretation of histologic sections is subjective, often extending, through a spectrum of possibilities”3 and that “the final morphologic diagnosis often is hedged with uncertainty.”4 In some circumstances, the hedges loom exceedingly tall, so tall that even though “we stand on the shoulders of giants,”5 we can scarcely see over them.It has been argued1 that rigorous scientific studies of the accuracy, reliability, and reproducibility of histopathologic diagnoses are rare but nonetheless disconcerting and that significant disagreement between pathologists exceeds 25% for difficult but clinically important diagnostic distinctions. With regard to malignant lymphomas, even the experts are seen to encounter difficulties: “It seems to me that pathologists are getting into difficulties in the area of non-Hodgkin's lymphoma. Not only do we clinicians have trouble in understanding pathologists, but at times they seem unable to understand each other.”6 Studies of diagnostic consensus for malignant lymphomas describe a level of agreement as low as 35% for practicing pathologists and 50% for experts applying the recently derived REAL classification. Attempts to predict prognosis by using histologic criteria for the grading of tumors are fraught with even greater difficulty.1 The fact that morphologic criteria still constitute the ‘gold standard’ for diagnosis in many conditions reflects not only the success of pathologists in coming to terms with subjectivity but also in some measure the lack of an alternative and more accurate diagnostic method. A serious and less well-recognized corollary is that in most cases we lack a reliable means of validating the surgical pathology diagnosis on which therapy is predicated.It may be argued that special stains evolved to meet just this need. The first special stains were variations of the basic biological dyes. Specific histochemical stains soon evolved based on the in situ identification of active enzymes within cells by the application of colorogenic substrates.7 Many histochemical methods were, however, technically exacting, both in terms of tissue preparation and staining protocol, a situation that led to recognition of the critical importance of appropriate positive and negative controls for interpretation of the findings.Although the methods were challenge enough, the reagents themselves were often poorly characterized and somewhat variable from manufacturer to manufacturer and lot to lot. In the United States, the Biologic Stain Commission (BSC) was founded in 1944 as a nonprofit corporation in large part to address the problem of standardization of chemical stains.8 The BSC achieved considerable success with regard to biological dyes, but with the advent of immunohistochemistry, special stain technology was about to take a new leap forward, with great opportunities and new challenges in standardization.In 1978 immunohistochemistry was only 4 years old. Nonetheless, imbued with the enthusiasm of relative youth and fortified by a mere 6 years of hands-on experience in the field, I was able to write of “the potential value of immunohistochemical methods as an immediate aid to diagnosis, and more importantly as a tool for the reshaping and redefining of current histologic criteria.”4 At that time, the whole field of immunohistochemistry as applied to routine formalin paraffin sections was encompassed in a score or so of manuscripts from around the world. By 1986 the literature had surpassed the 1000 mark, and I was moved to add, “Widespread application of immunohistological methods, by virtue of their great specificity, inevitably will transform histopathology from something resembling an art into something more closely resembling a science.”9 Although the utility of immunohistochemistry is accepted today by most pathologists, my optimism as to the scientific application of the method appears to have been somewhat misplaced in view of the lack of standardization and reproducibility among different laboratories. Immunohistochemistry has become an established method in surgical pathology—every (well, almost every) large academic center in the world uses the method routinely, some staining 10 000 or more slides per year. Immunohistochemistry is also offered by reference laboratories and even by many small community hospitals, where it is applied to the occasional case. But therein lies the rub; enormous variability has developed in terms of the reagents available, the detection methods used, and the interpretation and reporting of immunohistochemical findings. For these reasons, among others, the members of the BSC began, in the 1980s, to turn its attention to immunohistochemistry, sponsoring a series of workshops that were attended by pathologists, manufacturers, and representatives of the Food and Drug Administration.10,11Immunohistochemistry is technically complex, and no aspect of this complexity can be ignored, from the moment of collecting the specimen to issuance of the final report. In this context, anatomic pathologists should be prepared to follow the fine example of their colleagues in clinical pathology and should broaden their responsibility to cover all aspects of the immunohistochemical assay.1,12,13 The complete immunohistochemical stain (Table 1) begins with the identification of a need for the performance of immunohistochemical studies and progresses through procurement of the specimen, identification of the appropriate stain, selection of the proper reagents and protocol, and correct performance of the protocol with the appropriate controls. It concludes with an evaluation of the results and a written report. An attendant goal of long-term follow-up (outcome analysis) remains just that, a goal.Selection of the appropriate stain for a particular diagnostic problem should be an integral part of the process. Such a selection cannot be made in isolation from the clinical situation (diagnostic problem) on the one hand or the capabilities and limitations of the immunohistochemical laboratory on the other, including those limitations imposed by the experience of the staff and the type of tissue (frozen or fixed) available for study. The blunderbuss approach of ordering a panel of 10 or more stains on every suspected large cell anaplastic tumor can no longer be justified economically or academically. For each stain used in a particular case, the pathologist should ask, and answer, the question, “What will a positive/negative result add to the diagnosis?” If the answer is not known in the field of pathology or to the individual pathologist, then the stain should not be performed, because it will not add to the diagnostic equation.In the face of the diagnostic conundrum that cannot be resolved by orthodox morphologic criteria, the pathologist has 2 main options. The first and more common modus operandi is to select stains from a panel that, in the experience of the pathologist, have proven to be of value in a particular diagnostic area (eg, anaplastic tumors, malignant lymphoma). The second approach uses an algorithm with sequential panels of selected stains that introduces a certain degree of logic and thought into the process but may compromise the turnaround time due to sequential staining.13 Once the appropriate stains have been selected, the determinations of whether such stains can be performed successfully on the tissue available, whether the necessary reagents are at hand, and whether the detection methods are tried and true must be made; all these determinations will be readily apparent from review of the within-run (same day) and run-run (day-to-day) control records, which are an intrinsic part of an established quality assurance program. Although obvious, it can scarcely be overemphasized that without detailed knowledge of the performance characteristics of each primary antibody, it is not possible to interpret the final result, even in the presence of controls that perform as expected.There is also an important ongoing shift in emphasis in the use of immunohistochemical stains from a primary focus on cell and tissue markers as an aid to the recognition and classification of tumors to the demonstration of cell products, receptors, or oncogenes of possible prognostic value, and the identification of infectious agents in situ.13,14 The precision required for this latter approach places new and more exacting demands on our ability to perform immunohistochemistry or in situ hybridization in a reproducible and standardized manner.Aspects of tissue handling also cannot be ignored. Pathologists have sought to undo some of the adverse affects of formalin fixation either by the use of controlled enzymatic digestion or more recently by the antigen retrieval technique.13,15,16 The latter method (sometimes known as heat-induced epitope retrieval or as unmasking) subjects sections to microwave heating in the presence of a retrieval solution that may serve to stabilize or postfix the antigens present. Although vigorous heating in a microwave oven offends basic instincts, the method has contributed to the reproducibility of immunostaining by providing a more uniform presentation of antigens in tissue sections than is otherwise present following inconsistent fixation in formalin. However, one significant caveat has been issued,1,16 namely that “as different antigen retrieval approaches are explored and propagated, there is a danger that different laboratories each will adopt a different procedure, producing varying degrees of restoration of antigenicity, that will add yet another variable to the overall process.” Unfortunately this warning has come to pass.16 Clearly, the preference is for anatomic pathologists to adopt more uniform and rigorous procedures for the fixation and processing of tissues, extending from the moment that the specimen is removed from the body to the end of the embedding process, particularly controlling the total time in fixative and the composition of the fixative itself. The plea also was made that “if we cannot have uniform fixation, let us at least strive towards a more uniform approach to ‘unfixation,’ utilizing standardized and well described antigen retrieval procedures.”1 The pressure for consistency in staining has grown even further because of the increasing focus on prognostic markers, where greater stringency is required.Again, in the clinical laboratory, we collect a specimen in the right tube with the right preservative and the right anticoagulant, and we process it in the prescribed manner, all without thinking. In anatomic pathology, we largely fail to accomplish any of the analogous steps, again without thinking.The National Committee for Clinical Laboratory Standards currently is developing detailed guidelines that address the issues of fixation, reagent selection, choice of protocol, and other aspects of the complete immunohistochemistry test. The document is in the final stages of approval, and publication is expected within the year.Although it is perhaps not practical to attempt to standardize a single protocol across all laboratories, because of the great variability in specimen procurement, fixation, and processing, it certainly should be possible to standardize protocols within a single laboratory. The goal is to ensure run-to-run reproducibility, evaluated against standard control sections from day to day and week to week. Achieving this goal requires careful selection of reagents, rigorous use of controls, and strict adherence to the protocol for preparation of slides, application of reagents to slides, and incubation times. Today, it is increasingly clear that, just as with complex procedures in the clinical laboratory, consistency of performance can only be achieved by automation of the score or more separate steps that constitute an immunohistochemical stain.1 The Model T Ford immunostainer pioneered by David Brigati more than a decade ago has given way to a wide range of automated immunostainers, the best of which will outperform even the best technologist in terms of consistency during a sustained period. Of course, such instruments should not be viewed as a replacement for a skilled technologist but rather as an adjunct technology that frees the technologist to focus on issues of quality assurance and interpretation.The issue of selection of appropriate high-quality antibodies and reagents was addressed directly by the BSC in conjunction with the Food and Drug Administration. The result was the publication of a set of guidelines for package inserts.11 This document suggested not only a standard outline for manufacturers to follow in the testing and marketing of reagents but also provided a series of recommendations for positive and negative controls that could be followed by both manufacturers and performing laboratories (Table 2). By adherence to these guidelines, manufacturers will provide pathologists with more detailed information on the specificity and working conditions of antibodies than currently is available.With regard to detection procedures, the BSC concluded that it was unable to recommend a standardized staining protocol from the many variations in current use. Establishing a single universal protocol would seem to be impossible, recognizing that the different laboratories have different needs, and probably is not desirable, providing that all laboratories use appropriate controls. Here again is a spin-off advantage derived from the adoption of automated methods in that, just as in the clinical laboratory, automation tends to force compliance in terms of reagent choice and protocol, placing the onus of reagent qualification on the manufacturer, with a resulting increase in quality of product from the better manufacturers. Conversely, pathologists should recognize that any departure from the manufacturer's protocol or reagents effectively voids the responsibility of the manufacturer and places responsibility for validation totally on the performing laboratory.The experience and training of the technologist who is performing the stain clearly are critical to this process. Again, following the model of the clinical laboratory, one major factor in ensuring reproducibility may be the growth of automation, with the extended capability for consistency and control that is inherent in the automated process. The availability of automated immunostainers has proven of overall benefit in many smaller laboratories, if not some larger ones, that do not have the luxury of highly skilled staff who are experienced in reagent titration and quality control. It is certainly our experience that the use of an automated stainer the reproducibility and of a wide of immunohistochemical other not all agree on the choice of hands-on of the that most closely need and are necessary a With any automated there is a training during which and difficulties may be However, as these are the end results far the quality that is by even the best technologist on a of Pathologists quality assurance as a of that all pathology in the of have been in a appropriate to quality assurance is more a set than a but it the more of quality which is the of and so derived as to and correct in the described pathology in the years in a it by the new and automated that the clinical laboratory. The laboratory performing only a stains the of quality assurance in and quality control in a in for a then it in it to the technologist and the pathologist may an stain, but applied to immunohistochemistry it is a for to the and reproducibility of immunohistochemistry, we would be well to take a from our colleagues in the clinical for antibodies do not from immunohistochemistry in any way in terms of or performance the end is somewhat as the degree of staining of certain tissue and also However, there are in stringency of of the end in that strict criteria for interpretation of in to reference that have been available for many protocols for quality and of the reagents are part of routine practice in the clinical laboratory, to a large degree by clinical laboratory that are intrinsic to but have on anatomic pathology only recently guidelines of the Laboratory of If we are to immunohistochemistry to a level of reproducibility that in the clinical laboratory, then we must immunohistochemistry to a degree of including all its and each of the of the BSC and of the for Immunohistochemistry used by during the laboratory process. The criteria for of an immunohistochemistry laboratory were from to producing a more set of for the performance of immunohistochemical Many aspects of this have been in the National Committee for Clinical Laboratory Standards guidelines, which should to all laboratories performing immunohistochemistry in a diagnostic In an Immunohistochemistry and in which laboratories paraffin sections of cases with to perform a panel of immunohistochemical stains and controls. performance of the the laboratories their to the where results are and A detailed is provided to each The as an of the performance and interpretation of immunohistochemical stains on it not to measure the of specimen handling and fixation procedures, a that can only be by quality control critical problem for immunohistochemistry has been the lack of universal controls or reference a problem was in the of quality assurance procedures in the clinical For it was by large standardized and these available to both manufacturers and laboratories. In established other of which were provided to the of laboratories in testing The results from these different laboratories to provide validation of these as reference An analogous for immunohistochemistry, but the of reference of characterized is of to serve as a available BSC has the of a reference standard or a series of that could be made available for However, as the solution is not so with regard to characterized paraffin that cannot be and are available in no 2 tumors are provide an reference standard could even be prepared from the of cases used in the providing that has been evaluated and by different laboratories. But such even though they of are not The BSC has the of developing of standard reference in the of of tissues, including tumor cell or tumor in with to a from the the of of and a for the and of an The of this are into to this would do to consistency and into the The interpretation and of the should be in the of the overall of the presence of positive or lack of course, a complex It is a of the performance and of the proper controls and the experience of the laboratory performing the the pathologist for the A of difficulties into this including some of the problems that in surgical pathology in is If we as a then do we of positive cells a positive many a second aspect to interpretation to the way we at an the of a particular set of staining results in to the diagnosis or prognosis of the Here one can do no better than to more to the established in the process of are that are more readily than The literature has far the written by 1986 to a that in a single year. surgical pathology immunohistochemical to some with the literature is a challenge that it is to meet in within some such as the literature on is still with another due in The literature can scarcely with the of the and and even if it the a can scarcely with the to about such as Immunohistochemistry and meet the need of that cannot The pioneered by may provide for the but from of and places great demands on the for of but not least in the complete immunohistochemical stain, there are the issues of and In these there is The of surgical pathology has been in certain critical (eg, of A of studies have that immunohistochemistry is In one of more than anaplastic the diagnosis of or was in 50% of cases following basic immunohistochemical providing of a to were in In a separate of it was concluded that immunohistochemistry contributed to the diagnosis in almost 50% and was in the In a of poorly cell or cell tumors that could not be on the of morphologic immunohistochemistry a diagnosis in of the and of the of difficult cases seen in around the including our experience at the of of 000 cases during have been of the even the for appropriate immunohistochemical must be viewed against this for an individual immunohistochemical stain are not for some the of laboratory and for interpretation by a pathologist must also be into A panel of 2 or immunohistochemical stains that a from a may of that may have been by extended or an the detection of by immunohistochemical methods in or of with potential to a large with significant criteria, as by for such procedures as and have yet to be developed for the currently available immunohistochemistry as a and that has become an part of surgical pathology with regard to standardization and quality with automation of the and perhaps even the of basic results by will certainly the value still of this written for immunohistochemistry, to in situ hybridization methods and reagents used the all laboratories that perform immunohistochemical staining for diagnostic are to in the Laboratory and A for reporting immunohistochemical results we would be to follow the National Committee for Clinical Laboratory Standards is expected to quality assurance guidelines for immunohistochemistry in the further to the for overall quality and these almost certainly will but to a degree they are to the The performance and practice of immunohistochemistry are aspects of the practice of It is the of the pathologist, and it is our as pathologists, we do not this either by choice or then we must to and of If this in immunohistochemistry, that we have been to the we who to have the and is are to for preparation of the