For Your Reference

Michael Schubert interviews Keith Miller

Have you ever wondered how accurate your immunostaining really is? It’s a subject of deep interest to everyone from pathologists to regulators – but, until now, the ability to verify that accuracy has been limited due to a lack of universally available reference standard controls. Without such standards, it can be difficult to determine test sensitivity, ensure consistency and repeatability, and compare results between laboratories. To address this gap, a new group – the Consortium for Analytic Standardization in Immunohistochemistry (CASI) – aims to identify objective, quantifiable analytic sensitivity guidelines for immunohistochemistry (IHC) assays and establish reference materials to help laboratories ensure accurate, consistent testing for all patients.

Can you describe CASI’s work?
 

The CASI team are working on developing and validating a series of standardized controls for some of the companion diagnostic immunohistochemistry (IHC) tests and other important diagnostic IHC antibodies. The individual controls are made up of measurable amounts of target antigen (in the form of a specific protein created for a specific antibody) added to microbeads. There will be a range of concentrations of the specific protein developed for each specific test.

At present, the CASI team is focusing on having four control spots for each target on a slide next to the test section. The specific protein concentration ranges are aimed at encompassing the entire detection range, from lower to upper limit. This is still early days, but some tests may also need a spot that should be beyond the lower limit of detection with an approved assay. If this control lights up – for example, with a HER2 assay – it should indicate that the patient test section in question may be overstained, warning the pathologist that there is a risk of over-calling the case. If, on the other hand, the spot whose protein concentration is at the upper limit of detection fails to stain properly, this will provide a warning that there is a risk of under-calling.

There is a lot of work going on with tissue microarrays for HER2 at present, taken from a range of breast cancer cases that exhibit a spectrum of HER2 staining. The study includes both IHC and ISH testing.

In addition to the huge amounts of IHC staining needed to help with the calibration of individual control spots, there is a project underway to develop software that can read the intensity of staining within each of the control spots. This software will operate within a digital slide scanner and should be compatible with most scanners. That’s the wish!

What prompted the group’s launch?
 

It was Steve Bogen’s excellent idea to launch the CASI group. Steve is an eminent biochemist in Boston. I don’t know how he discovered the lack of reference standards in the IHC world, but he visited me when I was lead for the UCL Advanced Diagnostics lab and UKNEQAS-ICC&ISH in London some 12 or more years ago to discuss the idea of using specific protein/peptide spots as standardized controls for IHC. My colleagues and I were absolutely delighted by the proposal – but the ability to develop his idea and use some sort of digital slide analysis was beyond us at the time. Just before I retired in 2018, Steve got in touch again to see if we could work together and develop this project. Microbeads had been added to his proposal by then and, with all the technological developments that had taken place since Steve first visited, I felt sure the tools were now available to get his project up and running. I suspect others thought the same.

In your opinion, what are the major analytical challenges currently facing IHC?
 

I will avoid going into detail about the issues with formalin fixation and paraffin sections because there are just too many. Instead, I will concentrate on the main issues with IHC.

Selecting appropriate controls for IHC is very problematic at times – especially if only biopsy material is available. FFPE tissue, even in blocks, does not retain its original antigenicity over time. Often, paraffin blocks are stored in inappropriate conditions, which can make the loss of target antigen even worse. Paraffin sections, too, can lose their antigenicity if not stored appropriately.

FFPE cell lines are available for many targets, but not all. For a lot of clinically important targets, there are no cell lines available with the appropriate range of IHC staining.

Manufacturers have difficulty at times in eliminating batch-to-batch variation that affects their companion diagnostic kits. This is simply because either the control material they use is deteriorating or they must source other FFPE blocks that have been fixed differently to the batches of controls they employed previously.

The commonly used automated IHC instruments are not able to incubate every slide with specific IHC reagents for exactly the same amount of time as specified by the onboard staining protocol. I was once informed by one eminent IHC instrument engineer in the field that this is because only a single robot delivers the main IHC reagents for up to 30 slides and it is not possible to ensure exact timings for all the different tests that can be on board at one time. These small timing differences between slides can affect the staining result. Add to this the batch variation of IHC staining kits and the fact that there are now targeted cancer therapies licensed for use when a relatively small amount of target antigen is detected and it is quite clear that the IHC staining variation problem can potentially be catastrophic for some cancer patients. 

To further explain, because many laboratories like to use controls that show relatively strong staining for a given target, it is possible that a weak positive case could become an undetected false negative. Target antigen-rich controls can often show good staining even if the assay concerned is a little less sensitive than normal, which can happen for a variety of reasons. This risk applies to tests such as HER2, PD-L1, and ALK.

Another problematic area associated with companion diagnostics can be interpretation, especially when small amounts of protein must be identified and converted to a percentage of tumor cells positive within a given section. Having an IHC quality baseline provided by reference standard controls might help the development of some wonderful AI digital slide technology that could assist with more complex interpretation.

I believe that CASI’s work will overcome many of the issues affecting the sourcing of IHC controls today. There will be those who say the protein spots will not provide information on localization of the target antigen with a given IHC assay, but adding an appropriate FFPE cell line can overcome this issue.

How should pathologists and analytical scientists interact to further the consortium’s goals?
 

Our first step is to get the message about CASI out there. Publications alone will probably not get the message across at first – but we have a website in development that might also help.

My feeling is that we need to introduce the existence of CASI and its work at national and international pathology meetings before the various reference standard controls and related analytical software are ready for clinical use. I believe that, by doing this, we can highlight some of the problems that affect IHC testing today and explain why reference standards are needed.

UK pathologists are reliant on their scientists when it comes to understanding the weaknesses with IHC. Unfortunately, a lot of the older scientist who were around when IHC was developed into a significant diagnostic and predictive tool have now retired. Today, the younger generation are using push-button technologies and, in some cases, lack a thorough understanding of why an IHC test may go wrong. Worse still, a small number will not even recognize when a particular test result might be false. A lack of appropriate training is the main problem and this is unlikely to be rectified soon due to resource limitations. Having said this, the UK’s EQA Scheme, like other EQA programs around the world, is helping educate all in the field.

The other problem area that reference standard controls can potentially help eliminate is the batch variation affecting kits mentioned above. It will, of course, come down to how precise the CASI controls and related analytical software will be. Significantly reducing batch variation with HER2 IHC, for instance, could mean that less FISH/ISH testing will be required. I remember coming across several different batches of HER2 kits back in the day at our UCL-Advanced Diagnostics Laboratory when the FISH request rate went through the roof. In one case, we found many 2+ cases that were FISH amplified, suggesting that those kits were a little on the weak side; we also found batches where the number of HER2 FISH amplification cases dropped like a stone, indicating that those kits were probably too strong. Essentially, it seemed that a number of what would have been 2+ cases were probably in the 3+ category due to the oversensitive kit. I must also mention we were receiving several hundred cases for HER2 testing a month at the time, so we were confident that the significant changes in FISH positive rates were not due to natural variation.

It will take time to get the CASI controls into laboratories. If, as I expect, the combination of reference slides and analytical software help improve IHC standardization, I am confident that IHC kit manufacturers will see the benefit of using the system. Once this happens, more laboratories will likely adopt the CASI approach, too.

QA/QC is often overlooked or deprioritized in busy labs – how will the consortium’s work support improvements?
 

My wish, as a director of an EQA scheme, was to send out standardized controls for certain IHC targets that could be stained by the participant laboratory, then scanned with the lab’s digital scanner, uploaded to our website, and assessed automatically using AI by our EQA scheme. For me, the CASI work brings this wish another step closer to reality.

Of course, there will still be a need to assess the same type of IHC stain on tissue sections prepared in in the participant laboratory. For now, though, this will have to remain a human eye process.

Nevertheless, the CASI system can only add greater confidence to the whole EQA process. If the system indicates that the IHC assay is working well and the participant sections are not showing the same, it’s much easier to highlight where the problems might be – with poor fixation being one of the main suspects!