What is happening to immunohistochemistry?

Immunohistochemistry is a commonly used technique in research and pathology laboratories worldwide. However, in recent years, there has been a significant decrease in the number of Pubmed entries using the term immunohistochemistry. This decline can be attributed to two factors: increased awareness of the issue of unreliable research antibodies and the availability of novel RNA in situ hybridization techniques. Using the example of immunohistochemistry, this text discusses the factors that can affect good laboratory and publishing practices, or their lack thereof.


INTRODUCTION
Immunohistochemistry, a technique invented in the mid-20th century to detect bacteria using antibodies obtained from immunized animals, has been widely used since the mid-1970s to map the distribution of proteins in cells and tissues, mark specific cell types, verify knockout strategies, and evaluate diseased tissues. The success of immunohistochemistry cannot be overstated, with $2.5 billion worth of research antibodies purchased in 2015 [1] , and 30 000 PubMed entries on the topic in 2011 ( Figure 1A). In comparison, western blot and flow cytometry, two other popular antibody-based techniques, accounted for only 7000 and 11 000 PubMed entries, respectively, that same year.
However, I was surprised to find that the number of PubMed entries for immunohistochemistry fell to 20 000 in 2021 ( Figure 1A).
Even more puzzling is that western blot and flow cytometry have not been affected by such a decline; in fact, PubMed entries for those techniques have continued to rise in recent years ( Figure 1A). One wonders if laboratories are either publishing less on immunohistochemistry or referring to it by other terms such as "immunostaining" and "immunofluorescence". However, the use of "immunostaining" has remained low and relatively constant for many years ( Figure 1B), and although the use of "immunofluorescence" is growing, it cannot explain the decline in immunohistochemistry. The only logical conclusion is that research laboratories are publishing less on immunohistochemistry. What could possibly be the cause(s) of this sudden decrease in popularity of what was once a widely used laboratory technique?
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. BioEssays published by Wiley Periodicals LLC.

ARE WE NOT ALL DISENCHANTED WITH BAD ANTIBODIES?
Let's begin by addressing the thorny issue of unreliable antibodies.
Many commercial research antibodies do not recognize the peptide they are directed against, and often non-specifically bind to other proteins. [1] It is hard to say what proportion of commercial antibodies are unspecific, but in the case of antibodies against receptors, it is undoubtedly most of them. [2] Consequently, the likelihood of data published with unvalidated antibodies being erroneous is high.
To correct this situation, there have been numerous public calls for action on the front of antibody validation, usage guidelines, identification, and cataloging. [3] Among the important steps that can be used to validate a research antibody, the following methods can be employed in increasing order of stringency: blocking with an immunizing peptide, conducting a western blot on the tissue of interest, and validating specificity in a knockout model. However, very few academic journals require proof of proper antibody validation, and given how complex, time-consuming, and costly the process of antibody validation is, researchers have little incentive to validate their research antibodies. [1] Despite the pessimistic situation described above, the number of PubMed entries specifically focused on the issue of antibody validation has been on the rise in the last decade ( Figure 1C leading to speculation that laboratories may be more reluctant to publish data obtained with potentially unreliable antibodies. However, publications with flow cytometry and western blot are on the rise rather than declining ( Figure 1B), suggesting that concerns about antibodies are not the main reason for the decline in immunohistochemistry. While western blot is sometimes considered more reliable than immunohistochemistry, as one can easily tell the difference between unreliable antibodies by the molecular size of detected bands, there are numerous documented cases of antibodies that do not pass the knockout test, despite detecting a single band of the correct molecular size. [4] Similarly, unspecific binding of antibodies remains a common issue when performing flow cytometry. [5] For instance, antibody unpredicted cross-reactivity with multiple epitopes can generate false results. Yet only the immunohistochemistry literature is declining.
An additional piece of evidence indicating that concerns about antibodies are unlikely to be the sole cause of the decline in immunohistochemistry comes from the field of cell line biology. Misidentification of cell lines has been a recognized problem since the late 1960s. [6] Despite this knowledge, laboratories continue to publish research with misidentified cell lines, and according to the editors of the International Journal of Cancer, misidentification remains a common issue. [6] Some cell lines, such as Hep2, that were discovered to be misidentified decades ago continue to be widely used and published. Even worse, approximately 71% of manuscripts rejected by the International Journal of Cancer for cell line issues were eventually published in other journals with less stringent policies, despite the authors being made aware of the lack of validity of their data. [6] While initiatives such as stringent editorial policies, reagents catalogues, education about laboratory practices, and data sharing guidelines are useful in pro-moting good practices, they are not very effective at preventing the publication of erroneous data. Therefore, while scientists are becoming increasingly frustrated with unreliable antibodies, there is little evidence to suggest that this factor alone can explain the decline of immunohistochemistry.

Experimental biologists have a complex relationship with technology.
Laboratory techniques can become victims of fads and fashions, with the added value of a novel technique sometimes being oversold. As previously discussed, certain methods enjoy lasting popularity despite their shortcomings, while others are suddenly abandoned in favor of a newly developed technique that is perceived as more powerful, affordable, and easy to use. It is important to note that a technique may never gain popularity if it is too difficult to implement in the lab. The Falck-Hillarp technique to detect amines, for example, was key in starting the field of neurochemistry but was complicated and quickly replaced by immunohistochemistry. Sometimes, a technique can be forgotten and rediscovered much later, as was the case with light-sheet microscopy, which was invented over 100 years ago but seldom heard about before experiencing a revival in recent years. [7] Interestingly, there are alternative techniques to immunohistochemistry, such as RNA in situ hybridization, that can be used to map specific cell types or verify changes in gene expression. The main advantage of RNA in situ hybridization is that it uses probes that are complementary to the sequence of a target RNA transcript, which are controlled in design and synthesis unlike antibodies. However, RNA in situ hybridization is a laborious technique that requires specialized expertise, making it difficult for many laboratories to use it routinely, despite being invented in the late 1970s. Nevertheless, commercial kits such as RNAScope technology have made it easier for labs with little expertise in histology to perform RNA in situ hybridization effectively using novel "branched" probes that hybridize to specific RNA targets with minimal background. [8] Moreover, in the case of RNAScope technology, its sensitivity is known to be exceptionally high due to the utilization of branched probes and amplification detection processes. Moreover, RNAScope probes can be meticulously tailored for exceptional specificity, enabling the identification of genes within paralogous groups that possess remarkably high sequence similarity, surpassing the discriminatory capabilities of antibodies. Interestingly, these RNA-based techniques were introduced around 2010, which is the same time that immunohistochemistry entries began to decline.
Since 2010, over 7000 articles using RNAScope have been published ( Figure 1D). These numbers suggest that the rise of RNAscope largely compensates for the deficit of approximately 10 000 immunohistochemistry publications. Therefore, a new approach may be replacing immunohistochemistry as a technique of choice for many laboratories that were previously dissatisfied with unreliable antibodies. Lastly, the emergence of advanced sequencing techniques and gene expression studies has resulted in a surge of specific candidates, for which validated antibodies are currently unavailable. That this is not due to researchers' reluctance to utilize subpar antibodies; rather, it stems from the unavailability of suitable antibodies for experimental use.

CONCLUSION
The recent decline in immunohistochemistry publications can be attributed to a combination of two main factors: a critical mass of researchers being frustrated with antibodies, combined with the availability of a novel approach that allows RNA in situ hybridization with relative ease. Several lessons can be drawn from the decline of immunohistochemistry for anyone concerned with good laboratory and publishing practices.
Firstly, the fate of immunohistochemistry teaches us that raising concerns about the flaws of any technique does not automatically make researchers abandon it. We live in an era of productivity incentives, where there is pressure to constantly produce a large amount of "sexy" data at all costs. This partly explains why scientists have produced so much invalid immunohistochemistry for so long.
Secondly, all techniques in biology eventually fall out of favor and are replaced by novel approaches judged to be more powerful and/or practical. It appears that this is happening to immunohistochemistry.
While we may continue to see a decline in immunohistochemistry publications, the technique is unlikely to become completely obsolete. This is because immunohistochemistry does not provide the same information as RNA in situ hybridization, as the intracellular distribution and abundance of any given mRNA transcript may differ from that of the peptide it encodes. Moreover, immunohistochemistry can obviously generate high quality data when properly executed using validated antibodies. It will remain a useful technique for diagnostic and prognostic purposes in the field of anatomical pathology. Moreover, the possibility of a resurgence in the popularity of immunohistochemistry should not be dismissed. It is indeed true that immunohistochemistry has made notable advancements in multiplexing techniques, as well as benefiting from artificial intelligence-assisted analysis, to name a few. However, it remains uncertain whether these advancements alone would be sufficient to address the challenge of unreliable antibodies.
Lastly, no matter how powerful a novel technique may be, perfect techniques do not exist. For example, RNA in situ hybridization suffers from issues that can invalidate its validity, such as suboptimal tissue preparation, lipofuscin artifacts, and a lack of rigorous imaging protocols. Without due diligence to experimental controls, any laboratory technique can produce invalid results. Therefore, it would be premature to celebrate the rise of RNAScope as the end of invalid histology.
Scientists are condemned to constantly validate their methods and reagents, akin to a Sisyphean struggle.