Quality Assurance in Immunohistochemistry in the Age of Precision Medicine

What is This Chapter About?

This chapter summarizes our new understanding of IHC as an evolving method of testing and keys in on the most relevant concepts that have been brought about after much struggle with previous approaches, many of which failed to deliver the level of results that precision medicine demands. ^, It also provides guidance for terminology and definitions that will enable pathologists, laboratory technologists, oncologists, members of the pharmaceutical industry, and manufacturers of laboratory diagnostic equipment/reagents, to use the “same language” and to come to terms with the strengths and limitations of this methodology. For example, the term “biomarker” may be used by some pathologists only for those IHC assays that are predictive. However, “biomarker” is any characteristic that is objectively assessed (measured, detected, evaluated) as an indicator of biological or pathologic processes, or of pharmacologic responses to a therapeutic intervention. Hence, a “biomarker assay” is not a “biomarker”; rather, it is a methodology that can be applied to objectively analyze/characterize a “biomarker.”

This chapter focuses on QA, especially on validation of type 2 IHC predictive biomarker assays, namely those that are used to determine eligibility for specific drugs in specific disease settings.

Who is the Target Audience for This Chapter?

Although many of the concepts and terms depicted in this chapter may sound “technical,” they are of critical importance for the safe application of IHC to clinical practice, and hence, the expectation is that pathologists will be well acquainted with them. Since the complexities of biomarker testing extends into pre-analytical and post-analytical phases, it is also expected that we must highlight the importance of QA for predictive biomarkers to surgeons, oncologists, and other colleagues that are involved in processes that have relevance to and impact on, the results of IHC biomarker assays.

“Immunohistochemistry” and “Immunocytochemistry”

IHC, as a laboratory method/test, is an in situ qualitative test that detects specific proteins by using antibodies. It is performed on formalin-fixed paraffin embedded (FFPE) tissue or frozen sections, where a chemical reaction results in chromogen deposition at the sites of specific protein-primary antibody binding, and results are generated (i.e., read out) by pathologists +/− image analysis (IA). “Immunocytochemistry” is often used synonymously, but it reasonable to suggest the term be limited to those assays that are not performed on the tissue sections, but rather on cytology slides such as imprints (touch preparations) and smears (fine needle aspiration smears, peripheral blood or bone marrow smears, cytospins, etc.). ^, FFPE cell blocks represent a grey zone depending on whether they contain small fragments of tissue, cells only, or both.

Test Versus Immunohistochemistry Assay Versus Immunohistochemistry Protocol Versus Antibody

Historically, the terms “test,” “assay,” “protocol,” and even “antibody” were used interchangeably in published literature and in internal clinical laboratory documents. However, it is important to clearly distinguish between them in order to minimize the potential for lexical ambiguity, which may have either direct or indirect impact on the interpretation of validation requirements.

Test—IHC is a test; fluorescence in situ hybridization (FISH) is a test; PCR is a test. A laboratory “test” is a method used for analyzing analytes that can be assessed/detected using that method. For example, IHC can be used to detect proteins in situ, but not RNA or DNA. Therefore, we cannot test for RNA or DNA by IHC. Admittedly, it is common for “test” and “assay” to be used interchangeably. However, due to the rising complexity of biomarker testing in precision medicine, it may be time to make a commitment to separate the two.

IHC Assay—While a test is a methodology used (e.g., IHC, FISH, other), an assay is a laboratory-based qualitative or quantitative analysis for the presence, amount, and/or functional activity of a specific analyte of interest that includes all phases required for the analysis in accordance with how this term is used colloquially by the US Food and Drug Administration (FDA). Every assay has defined pre-analytical, analytical, and post-analytical conditions that are prerequisites for testing. There may be many assays for a single biomarker with the same or different purposes that may use one or more methodologies (e.g., IHC and/or FISH). Assays may be approved by regulatory agencies or they may be developed by an IHC laboratory, so-called “laboratory developed tests” (LDTs). It is worth noting that within this definitional framework, LDT is a misnomer as it is really a “laboratory developed assay” (LDA); however, it is likely that use of the historical designation “LDT” will continue.

IHC Protocol—A protocol is a set of written instructions on how to perform a qualitative or quantitative laboratory analysis for the presence, amount, or functional activity of an analyte. In other words, a protocol indicates how an assay should be performed by the laboratory. An IHC protocol specifies conditions of the analytical phase including the reagents (e.g., which primary antibody, antigen retrieval buffers, detection system, etc., to use), instruments (e.g., the specific automated stainer, no automated stainer/manual, etc.), and conditions (e.g., time, temperature, without or with amplification, with or without enhancement, etc.). The three critical components of the IHC protocol (primary antibody, antigen retrieval, detection/visualization) are illustrated in Fig. 2.1 .

Antibody—Pathologists almost universally use the term “antibody” instead of either “IHC assay” or “IHC biomarker assay.” This has led to wide-spread underappreciation of the relevance of the IHC protocol and IHC readout to IHC assays. It also makes it difficult to appreciate the importance of purpose (see below) in IHC biomarker assays and leads to the erroneous assumption that having validated one IHC assay using a primary antibody raised against an epitope of a particular molecule in a particular disease condition (or tumor type), means that the developed assay will be valid for all IHC testing for that molecule irrespective of the tumor/tissue type. The SP142 rabbit monoclonal antibody used in the VENTANA PD-L1 (SP142) Assay is often referred to as having an analytical sensitivity much lower than that of the PD-L1 SP263 “antibody,” or the mouse monoclonal PD-L1 antibodies 22C3 and 28-8. ^, These published results were generated by using an FDA-approved CDx Kit with an IHC protocol that was developed for the needs of specific clinical trial(s). The question is whether it is possible to increase the analytical sensitivity of an assay using the SP142 PD-L1 antibody clone, by developing a more sensitive IHC protocol? If not, then the antibody really may have low sensitivity for PD-L1. However, if the answer is yes, then it may be that the apparent low sensitivity was by design. Indeed, when a more sensitive LDT IHC protocol was developed using the SP142 clone, the analytical sensitivity on the same tumor samples radically changed and became quite similar to that of the DAKO PD-L1 IHC 22C3 pharmDx ( Fig. 2.2 ).

In summary, if one were to be asked which test could be used for a predictive biomarker for immunotherapy, the answer could be “IHC,” “NGS,” or other. If the answer was “IHC,” then a further question could be to ask which IHC assay may be used for selecting patients with gastric cancer for treatment with pembrolizumab, where the answer could be “the DAKO PD-L1 IHC 22C3 pharmDx” or “an in-house PD-L1 LDT biomarker assay using the SP263 primary antibody that was specifically validated for that purpose.” One could then ask for details of the DAKO PD-L1 IHC 22C3 pharmDx protocol that is provided in the product specification sheet, or for details of the PD-L1 LDT internally validated protocol, including which antibody was used in that IHC protocol.

Biomarker Versus Immunohistochemistry Biomarker Assay Versus Immunohistochemistry Assay

In IHC, the difference between a “biomarker” and an “assay for a biomarker” or “biomarker assay” is essentially the same as for any other testing method. The National Institutes of Health (NIH) Working Group definition of a “biomarker” is as follows: a characteristic that is objectively measured and/or evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention. A “test” for a biomarker is a testing methodology used to objectively measure and/or evaluate the biomarker. In this case, it would be IHC. An assay, being a testing procedure that includes all phases of testing, is an IHC testing procedure as specified by the manufacturer (usually in the specification sheet) or by the laboratory standard operating procedures (SOPs). There are diagnostic, prognostic, predictive, and surrogate endpoint biomarkers with various assays that use different tests (methodologies) to assess them. An IHC assay becomes an IHC biomarker assay when there is evidence that its results objectively indicate that a normal biological process, a pathogenic process, or a pharmacologic response is informative for some aspect of a therapeutic intervention. Therefore, an IHC assay becomes an IHC biomarker assay when it is validated for a specific purpose. In other words, an IHC biomarker assay is a fit-for-purpose IHC assay. This applies not only to predictive IHC biomarker assays but also equally to diagnostic and prognostic IHC biomarker assays, thus indicating that all IHC assays used in clinical practice require purpose-based validation.

For example, trisomy 21 is an excellent diagnostic biomarker for Down syndrome and was also shown to be an adverse prognostic biomarker in follicular lymphoma. ^, However, there are different biomarker tests (methodologies) for detecting trisomy 21 (e.g., karyotyping, FISH) that have been incorporated into different assays (including pre-analytical, analytical, and post-analytical conditions as defined by laboratory SOPs). The sensitivity and specificity of these various tests may be different for the diagnosis of Down Syndrome (one purpose) as well as for prognosis in follicular lymphoma (another purpose); these differences may also exist between separate laboratories performing a similar assay depending on the exact laboratory protocols used. For IHC, the example of PD-L1 is particularly important. PD-L1 is a biomarker in the context of immunotherapy. An IHC assay using the PD-L1 E1L3N antibody clone developed for optimal signal-to-noise ratio is a PD-L1 IHC assay. The DAKO IHC 22C3 pharmDx for selecting patients with NSCLC for treatment with pembrolizumab, or the VENTANA PD-L1 (SP142) Assay for selecting patients with TNBC for treatment with atezolizumab + chemo, are both PD-L1 IHC biomarker assays (CDx Kit). Also, a PD-L1 assay using the E1L3N clone that has been validated as a surrogate assay for the DAKO IHC 22C3 pharmDx for selecting patients with NSCLC for treatment with pembrolizumab is also a PD-L1 IHC biomarker assay (LDT). Therefore, it is essential to distinguish between the biomarker, the IHC assay, and the IHC biomarker assay as this has direct impact on the our approach to the validation of IHC biomarker assays, especially those used to determine patient eligibility for targeted therapies. Clinical IHC laboratories are required to validate (or verify) IHC biomarker assays, not the biomarkers themselves.

Type 1 Immunohistochemistry Assay Versus Type 2 Immunohistochemistry Assay

Regulatory agencies such as the FDA use risk-based classification for IHC assays that focuses on the purpose of the assays (e.g., Class 1, 2, and 3). However, this approach does not necessarily have direct applicability to clinical practice. In routine practice, it is how pathologists use the IHC assays that is most relevant, and it is entirely possible that pathologists do not always align the use of an IHC biomarker assay with the FDA-approved purpose. From the pathologist perspective, it is important to differentiate between IHC assays (1) that are intended to be diagnostic, where the results will be used by pathologists to determine cell differentiation, render diagnoses, and classify diseases, from (2) those that are still intended to be reported by pathologists, but where the end user is instead, the treating physician (most commonly, an oncologist). The former is a type 1 IHC assay; the latter is a type 2 IHC assay. In older literature, these were previously referred to as Class 1 and Class 2 IHC assays. However, “class” has since been replaced by “type” to avoid conflation with “class” as used by the FDA and other regulatory agencies. The current terminology of “Type 1 IHC assay” and “Type 2 IHC assay” is a QA-based classification that is more suitably aligned with modern laboratory practice.

Accuracy is the Measure of Quality for Immunohistochemistry Biomarker Assays

QA is an intrinsic component of laboratory testing. The principal goal of the totality of the numerous processes that represent laboratory QA is to ensure that the results generated by that laboratory’s assays are accurate/correct (specific and sensitive), every time (reproducible), and relevant to patient care (cover relevant reportable range and are reported in accordance with the purpose for which the assay was performed). Although these QA parameters were always applicable to IHC testing, it was only when IHC started to be significantly incorporated into the precision medicine framework (through the provision of predictive biomarker assays) that the need for a more modern approach to QA in the IHC laboratory was necessary not only for IHC laboratory technologists, but also for pathologists, as well as for oncologists. Pathologists are highly trained expert physicians and are typically able to discern when a diagnostic IHC biomarker assay is not working optimally because they draw on their knowledge of pathophysiology, the effects of disease processes on tissue morphology, and the results of other diagnostic biomarker assays (IHC +/− histochemistry or other); however, this is rarely possible for predictive IHC biomarker assays. Therefore, predictive IHC biomarker assays require a higher level of QA practice both in their development, as well as in the ongoing monitoring of their performance.

Now we turn our attention to the meaning of “accuracy” for a test that is qualitative in nature. If the purpose of an IHC assay is to determine the lineage of the lesional cells (e.g., cytoplasmic expression of myeloperoxidase [MPO] defines myeloid differentiation), then we what we want is a sensitive and specific assay that will stain only the cells that express MPO. We expect to see that the cytoplasm of myeloid cells is positive even when MPO is weakly expressed, and we would not tolerate any false-positive cytoplasmic reaction in non-myeloid cells. If the purpose of an IHC assay is to determine whether a patient with adenocarcinoma of the lung is eligible for targeted therapy with anaplastic lymphoma kinase (ALK) inhibitors, then we would want an IHC assay that has very high sensitivity and specificity for identifying such patients by showing cytoplasmic granular diffuse staining for ALK protein, in tumor cells with ALK rearrangements, even in cases that only weakly express the ALK protein. If the MPO and ALK assays perform as just described, then their quality is high.

For predictive biomarker assays, it is far more important that they be able to accurately stratify patients as close as possible to what was achieved in clinical trial(s), than that they “look good” (i.e., have excellent signal-to-noise ratio; see “Aesthetics vs. Accuracy” below). It follows, then, that for validation purposes it is necessary to determine/calculate an assay’s sensitivity and specificity as well as positive predictive value and negative predictive value. From a statistics perspective, this further leads to the need to have an adequately powered validation set (number of cases); assessment of analytical performance is not sufficient if IHC biomarker assay accuracy is not properly addressed. ^,

It is suggested that biomarker assays, which have a combined sensitivity and specificity that total at least 170, are likely to be clinically useful. An IHC biomarker assay with 95% sensitivity and 95% specificity (total of 190) can be considered an excellent assay. An assay that has 90% sensitivity and 90% specificity (total of 180) can be considered a good assay.

In summary, the quality of IHC is measured by the accuracy/correctness of the results as defined by the purpose for which the respective IHC assay was developed. ^{, ,}

Aesthetics Versus Accuracy

It may appear that an emphasis on accuracy regarding the quality of IHC testing excludes other important aspects of IHC quality such as good signal-to-noise ratio, good hematoxylin counterstaining, and good morphology. Although “nice, clean staining” is always helpful to pathologists reading the stained slides, it is far more important that the overall generated results are accurate, rather than aesthetically pleasing. Therefore, even when hematoxylin counterstaining is either too light or too dark or there is some background staining, if the reported results are accurate (i.e., can properly stratify patients for targeted or other therapies), then the IHC biomarker assay is of high quality. It is typical for excellent “signal-to-noise” ratio and accuracy of an IHC assay to go together; however, clinical practice has shown a sufficient number of exceptions such that clinical IHC laboratories and external quality assurance (EQA) proficiency testing (PT) providers should take note and adjust their internal processes to ensure that assessments of the quality of IHC assays be based on their accuracy for specific purposes in the first place. While this approach is more applicable to predictive IHC biomarker assays, it clearly also applies to any IHC assay. ^,