Regulatory Forum Opinion Piece*

In the context of monoclonal antibody drug development, tissue cross-reactivity (TCR) studies are immunohistochemical (IHC) studies conducted on a broad panel of normal body tissues, using the antibody therapeutic as the primary reagent. The ICH S6 guidance states that these studies are performed to support investigational new drug (IND) or clinical trial applications (CTAs) for the initial dosing of human subjects for all antibody-based therapeutics. The purpose of TCR studies is, ostensibly, to inform investigators and regulators of potential target organs of toxicity based upon the binding of the therapeutic agent to tissue components. The value of TCR studies in predicting toxicity that would not otherwise have been detected in in vivo pharmacology and toxicology studies in pharmacologically relevant animal models has been a subject of debate for more than a decade.

The history, challenges, and proper and improper uses of TCR studies were reviewed in this forum by Leach et al. in 2010. Coincident with this publication, a cross-industry survey on the utility of TCR assays in predicting human toxicity was separately published by a working group of the preclinical safety committee of the Biotechnology Industry Organization (BIO Safe) by Bussiere et al. (2011). Readers are referred to these excellent reviews for a detailed history of the TCR study.

The focus of this commentary is on the interpretation and reporting of TCR studies and the need to simplify them. In the conduct and interpretation of a TCR study, simply put, there are three potential outcomes for any given tissue: (1) negative staining, (2) expected positive staining, and (3) unexpected positive staining. Expected staining is seen in those tissues and cell types where the target is known to be expressed based on existing knowledge of target distribution. Identifying areas of unexpected staining, potentially alerting investigators of an in vivo target organ, is the raison d’être for the TCR study. Whether the unexpected staining represents a previously unknown site of target expression, or staining of a similar epitope on an “off target” molecule, is typically never ascertained. As in any histology-based assay, the interpretation is highly dependent on the opinion, experience, and skill of the observer who is usually (but not always) a pathologist. Regardless of who is interpreting the staining outcome, accurate interpretation of staining absolutely requires a sound experimental IHC method, and comparison of staining in investigational tissues with that in the positive and negative controls. To date, there has been little focus on how staining interpretation in TCR studies is determined and reported. This issue either reflects certain assumptions about how these studies are conducted or is akin to the old metaphor that no one wants to see how a sausage (or legislation) is made. When a TCR study is complete and neatly reported in narrative and tabular form, it generally looks very straightforward. But behind the scenes, debates occur commonly regarding appropriate due diligence for developing a good positive control for these assays and the criteria for positive tissue staining. These discussions, often in the context of peer review, are critical to optimal reporting that allows for useful risk assessment.

The technical challenges of developing a sound IHC method for these studies are many, not the least of which is the use of frozen tissue sections; the frozen tissues from the human donors are derived from biopsy or autopsy material and are frequently of uncertain quality and morphology. In addition, the use of therapeutic antibody formulations that are not intended to be used for IHC, with binding characteristics that can be further altered by labeling, presents a considerable challenge for some antibodies. In addition, positive staining for many targets is seldom clear-cut. Different laboratories and different pathologists or histologists have, not unexpectedly, a wide range of approaches, opinions, and criteria concerning what constitute a satisfactory positive control, and in turn, what constitute positive tissue staining. In some circles, development of a good staining protocol has been referred to as “witchcraft”—a description that does not exactly inspire confidence for a regulatory nonclinical safety study.

There is a need to simplify interpretation and reporting in these assays. The following thoughts regarding positive controls and the assessment of positive staining are proposed, with the aim of making what are often very complex, arduous, and confusing studies a little simpler, more definitive, and credible:

1. For tissue-bound targets (tissue receptors or other cell surface molecules), the only positive control that matters is a positive control tissue section.

   In IHC, positive control tissues are usually the normal or diseased organs or other pathologic tissue (i.e., tumors, inflamed tissue, etc.) where the target is thought to be expressed or overexpressed. These tissues represent a logical starting place for development of an IHC method. However, in TCR studies when staining fails in the expected positive control tissues and in an effort to demonstrate due diligence and complete the study, other types of in vitro samples are used to demonstrate positive staining. Examples include purified protein “spots” on glass slides for microscopy, Western blots, antigen-coated sepharose beads, or cytologic preparations (cytospins) of transfected overexpressing cell lines. While it is usually, if not always, possible to achieve a positive staining in one or more of these alternate in vitro environments, these samples do not represent the complex tissue microenvironment that is present in an ex vivo histologic tissue sample. As such, they provide no information as to the optimal staining conditions needed to stain the known/expected epitope in tissue. Demonstrating positive staining using these in vitro samples does not inform the observer of what positive staining might look like in the matrix of the tissue sample, and in this sense, these samples cannot truly represent useful positive controls.

   Additionally, if and when a positive control tissue is identified, the control staining must be repeated and unequivocal with every staining run (these studies typically take multiple staining runs because of the large number of tissues from multiple donors that need to be stained). If the positive control does not unequivocally work in a given staining run, that entire run should be discarded and staining repeated.

   To review, if a convincing, repeatable, positive control histologic tissue section cannot be identified during the methods-development phase of IHC, then the TCR study should be considered “not feasible” and the work considered complete. In this case, the detailed failed efforts to develop a sound IHC method for the TCR study should be documented in a report and submitted as part of the IND/CTA.

2. For soluble targets (secreted proteins and soluble ligands), a positive control tissue section is still required.

   Soluble targets present additional challenges, and the principal obstacle is that positive staining would not be expected in any tissue—only in the serum or extracellular environment. Because the object of the TCR study is to look for unexpected staining in tissue, a positive control tissue still must be identified; protein spots, Western blots, or cytologic preparations of overexpressing cell lines are not acceptable.

   There might be several ways of accomplishing this objective. One could xenograft an immortalized overexpressing tumor cell line in an immunodeficient mouse and collect the tumor tissue as a possible positive control sample. Other methods might include perfusing overexpressing cell lines or coated beads into fresh tissue (lung for instance) then using perfused lung tissue as a positive control. Other methods may also exist, but failure to develop a tissue positive control by one these methods should render the TCR study as “not feasible.”

3. The central criterion for reporting a tissue as “positive” is publication quality of the IHC.

   Overinterpretation is a big problem in TCR studies. Assuming that we have been successful in developing an IHC method in a positive control tissue, and before we interpret an investigational tissue as positive, we really need to ask ourselves, as pathologists, the following questions: “Am I willing to submit representative photomicrographs of this IHC experiment, including the positive and negative controls, to a peer-reviewed pathology journal as evidence of “positive” tissue staining?” Or alternatively, “Can I proudly stand up at an STP meeting and show the photomicrographs to all of my colleagues from around the world and declare this is a conclusive example of a positive stain, without fear of a long line of questioners lining up at the microphone?”

   In my estimation, staining should be obvious using the 10× objective (100× magnification), with a good signal–noise ratio, and clear staining of specific cells or a microanatomic region of an organ in order to pass muster. All other nonspecific or uninterpretable “staining” should be considered negative. We have all seen examples of nonspecific or uninterpretable staining including submicron-sized deposits of chromagen visible only under the 60× objective in a few cells of an organ (sometimes referred to as “+1” and/or “rare staining”). These would not pass the red face test as positive if presented in a public forum (a “threshold” problem). Alternatively, diffuse, smudgy, nonspecific staining without microanatomic localization (but not present in the negative controls) could be called positive by some, when, in fact, this staining pattern is more attributable to some biophysical characteristic of the primary antibody, or the tissue, or the procedure, and does not represent specific antigen–antibody interaction. To interpret this type of staining pattern as positive is a misleading overinterpretation and would generally not be accepted in a pathology/histology peer-reviewed journal as evidence of target expression or “positive” tissue staining.

Ultimately, TCR studies should be held to the same quality standard as any other IHC experiment, because that is what they are. As with any other pathology end point, interpretation requires professional judgment and always requires appropriate controls and peer review. The IHC method must produce convincing staining in the positive control tissue section (not an in vitro sample), and only convincing, repeatable tissue staining that would hold up under rigorous peer review for publication (or for credible presentation) should be called positive. Holding TCR studies to these standards would produce much better data, leading to better risk assessments, and produce some nice material for publication as well.

Please consider one final thought. There is still a need for a better understanding of value of this assay in predicting human toxicity that would not otherwise have been discovered in standard in vivo preclinical toxicology studies in pharmacologically relevant animal models. The existing publications and surveys cite examples of unexpected tissue staining that predicted a true clinical hazard, but these short, sanitized vignettes lack sufficient detail to be useful in excluding the possibility of coincidence instead of a cause–effect relationship. Good case studies are needed in the peer-reviewed literature to add credence to the idea that this assay helps safeguard patients in clinical trials from unexpected toxicities. These case reports would include photomicrographs of the unexpected human tissue staining in the TCR/IHC study; then provide descriptions of the toxicity encountered in clinical studies, which correlates with antibody staining of the target organ; and show a lack of correlative staining in animal tissues or in vivo toxicity in the animal studies. This would be welcome reading.