Improving Immunohistochemistry Standardization in your Laboratory
Dr. Farah Patell-Socha, Product Development Manager, Horizon Discovery
Dr. Vicky Spivey, Senior Scientist, Horizon Discovery
The information gained from IHC combined with microscopy provides a valuable perspective that can help to make sense of data obtained using other methods. Optimization and standardization of these variables are therefore needed to allow for consistent and reproducible results. Example variables include, but are not limited to: antigens, epitopes, sample preparation, fixation methods, antibodies and platforms
FP: Immunohistochemistry has evolved from an investigative tool to a widely used diagnostic technique. Although it has been employed for 50 years, the application of this method suffers from a lack of standardisation and reproducibility amongst different laboratories. IHC standardisation is indispensible for reliable and consistent results. Today we will be talking about variability arising from a typical IHC workflow and how we have developed for research use only genetically defined IHC reference standards as independent external controls for routine evaluation. These controls aim to address the issues around consistency and reproducibility of your assays. My colleague Dr Spivey, a senior scientist at Horizon will talk about how these standards can improve IHC standardisation and present data showing the concordance between cell lines and tissue staining results from the UK kin class email for ELK prebio scheme. 1.04
So before I begin I’d like you to think about this one point. What is the impact of assay failure in your lab and how do you monitor for it? Where can variability creep into a standard IHC workflow – this slide highlights a typical IHC workflow in a lab, starting from sample preparation to analysis. As you can see on the right hand side I’ve jotted a couple of key points that I will go through one by one. The first challenge starts right at the beginning of the process with differences in sample preparation, heterogeneity of the sample and fixing. Different labs follow different protocols in sample preparation and these can both influence ream results such as intensity of staining and tissue morphology. The second challenge relates to the thickness accuracy of the sections cut. We all know the difficulties and variations in section thicknesses that can result on different staining intensities and so therefore consistency is extremely important.
As I’m sure you all know when performing IHC assay labs use various staining it’s from different suppliers, antibodies from different vendors and automated staining platforms from different manufacturers and all can result in different staining results. Testing EML-4-ELK detection in FFPE for example, what antibody would you use? From our external studies we know that the labs are using a range of antibodies. B5F3 being the most popular ALK1 Dako clone being the most unpopular and the Leica antibody to name a few. And if the protocol is not optimised this can lead to false positives and negatives.
Next we come to the accuracy of quantification and interpretation. How accurate is your scoring method especially for biomarkers that you don’t yet have official scoring guidelines for. For labs in particular using quantitative digital pathology, what are the thresholds that you’re using to calibrate your assays? Finally what controls are you using to routinely monitor your assays? Is it easy to find? Is it a finite source? Is it consistent day on day, year on year? Over the next few slides we will discuss the challenges in more detail and show you some internal and external study data
So to resolve these challenges associated with standardisation, we developed genetically defined IHC reference standards with consistent protein levels for analytical validation and quantitative assessment for immunohistochemical assays. The aim was to create a standard that was renewable, consistent and reproducible for routine monitoring. This slide highlights a couple of its key features of the range of reference standards that we have manufacturers here at Horizon. We have developed a range of research use only reference standards for translocations, amplifications, SNP’s. Involved with the therapy. 4.33these standards come in a biomarker specific format as you can see on the left hand side, for EML-4-ELK, ROS1, MET and BRAF V600E. So these are biomarker specific slides, as well a multi-marker reference slide containing multiple markers for the specific tests for cancer – in this case we have developed a lung panel targeted ELK, ROS, EGFR and MET. These standards contain defined cores containing positive and negative protein expressing cell lines on the same slide. As you can see here the ELK reference standard slide a negative and a positive and core, whereas the MET and BRAF have a negative, intermediate and strong cell lines expressing different amounts of protein.
3 micron sections from these FFPE blocks are sectioned and mounted onto super frost positively charged slides. There is ample space to place your sample of choice and on slide control to add confidence to your IHC result for routine evaluation. Another unique feature of these standards is that the negative and positive cores contain extensively characterised cell lines molecular assays, IHC, western blot and FISH. For example, the negative cell lines in the ELK reference standard is not only negative by IHC but will also be negative by FISH, rtPCR and PCR etc. Therefore and similarly the positive cell line core on the ELK will also be positive by IHC, FISH, rtPCR etc. Therefore the negative and positive cell line cores that we use in our standards act as standards for the assessment of multiple diagnostic modalities.
We have also brought in quantitative digital pathology. We have used QDP for our IHC reference standard internal quality control. Biomarker specific algorithms are built for each biomarker and allow us to characterise and assess the negative, intermediate and strong protein expressing cell lines in our standards.
My colleague Dr Spivey, senior scientist at Horizon, has strong interest in the future of personalised medicine and has spoken on advances in diagnostic testing. He has been instrumental in developing of these FISH and IHC reference standards at Horizon. She will talk about how these standards have been manufactured and how they will improve IHC standardisation. She will also present data showing the concordance between cell lines and tissue staining from us in a UK class EML-4-ELK scheme.
VS: Thank you Farah. I’d like to start with how we manufacture these genetically defined standards. Let’s take EML-4-ELK as an example. We take a cell line wild type to the mutation of interest and a single cell; dilute it to create a clonal wild type cell line. Then using Horizon proprietary gene editing technology, the line is targeted to create a clonal mutant cell line which contains the mutation or translocation of interest, in this case an EML-4-ELK translocation. The result is you have a pair of isogenic cell lines which are highly characterised using SNP 6.0 to confirm cell line identity, Sanger sequencing to check that the engineering event was correct, digital PCR to check the allelic frequency and expression of the transcript and IHC and FISH to assess the staining pattern. In addition, as Farah introduced earlier we use for our IHC reference standards, quantitative digital pathology to evaluate the cell lines. All this results in a consistent, reproducible and renewable source of positive and negative reference material.
So using EML-4-ELK as an example, I’d like to start by presenting some data produced with our collaborators UK NEQAS and UCL Advanced diagnostics. As mentioned earlier, IHC variability can arise from different labs using different antibodies, protocols, platforms and tissue. We wanted to test our ELK standard alongside positive tissue to determine whether there is a concordance between cell line and tissue staining. In this study, NSCLC ALK positive tissue that was previously shown to be positive by FISH, was placed on Horizons EML-4-ALK reference standard slide and the slide was stained for ALK using Ventana benchmark system using the D5F3 antibody and the Optiview detection system. As you can see on the slide the negative cell line core shows no protein expression and the positive cell line core beneath was strongly positive with strong cytoplasmic staining detected in greater than 99% of cells using Imperio imaging analysis. Importantly the positive tissue stained to an equivalent level as the positiv4 cell line core demonstrating excellent concordance between tissue and cell line staining results. This data also matches results from additional laboratories that participated in our external evaluation programme.
Next, we now wanted to test how the ALK standard would perform on a different platform and detection system using the same D5F3 antibody. The EML-4-ALK reference standard slide was sent to another laboratory for ALK staining using the Dako link autostainer without the Optiview detection kit. The ALK reference standard worked as expected using both methods. Core A at the top containing the ALK wild type cell line was negative as expected whereas the cells with the ALK translocation core B contained cells that were 99% positive. This data highlights the variation between methods and detection kits and the importance of an external control for assay optimisation and monitoring.
The next objective was to use these genetically defined negative and positive ALK translocation cell lines in a proficiency testing scheme in collaboration with UKNEQAS. Data presented on this slide was generated by UK NEQAS as part of a pre pilot EQA scheme for ALK, and it had 36 participants. EML-4-ALK, their cell lines positive for ALK expression were mixed with the wild type cell lines 1:1 in an FFPE block. This slide highlights the concordance between the EML-4-ALK cell lines and tissue samples. I just walk you though the slide – sample D is the ALK negative cell line and sample E blow is the ALK negative tumour. Tissue sample E is negative but the macrophages are staining which is a good guide to the sensitivity of the IHC assay. A is the ALK positive cell line and beneath it is Sample F, the ALK positive tissue. In this example, samples were stained using the Roche D5F3 assay on the Ventana benchmark using the Optiview detection kit. In the table below most commonly used antibody is the Ventana Roche D5F3 clone which was used by 21 out of the 34 labs that provided feedback. An additional 7 antibodies were used by the participants and regardless of the platforms and antibodies used the defined ALK cell line performed as expected and in concordance with tissue samples. As mentioned earlier our objective was to develop genetically defined IHC reference standards with consistent protein expression levels for quantitative assessment of the IHC assays. We wanted to develop a tool that would allow us to achieve this goal. As Farah introduced earlier, analysis of IHC staining is very subjective. In addition to the molecular characterisation of the cell lines, we have adopted QDP to further define the positive and negative cell line cores. Taking EML-4-ALK as an example again, this reference standard containing the positive and negative cell lines was run on the Leica bond mac using our optimised protocol and the images were scanned using the Imperio image system. From these scanned images the ALK algorithm was developed. So the intensity of the individual cells in the cores were analysed the Visiopharm software shown here. And the resulting histograms were compared to create the optimum cut-off between positive and negative cells, minimising false positive and negatives results. The threshold shown here on the right were then set and used for all ongoing analysis for consistency and reproducibility. The H-score provides additional information about the staining as it is the score of staining intensity and can monitor changes more accurately than a simple positive or negative cut-off. Over the coming months we will have data showing the comparison between cell line and tissue post quantitative assessment.
To summarise the key points, we have produced characterised and genetically defined cell line standards. The standards show excellent concordance between cell line and tissue staining results and highlights variation between methods and detection kits, all demonstrating the need for external controls. In addition to the molecular characterisation we have adopted QDP to further define positive and negative cell line cores.
FP: Thanks Vicky for that. So to follow on from the data, I would like to briefly run through the format and application of our OC ref standards. The reference standards that we have created come in two different formats. We have the FFPE blocks and we have sections that re mounted on slides. As I mentioned before we have biomarker specific slides, with one marker with its negative and positive control and you’ve got the lung marker panel in this instance but in the future and coming months we will have the breast panel – we will have biomarkers such as ERPR, Ki67, DBL-1 very soon. So this will feed into these different pipelines. If you are interested in FFPE blocks, we have FFPE blocks as well.
So the applications are quite simple. These standards can be used for routine controls for an IHC and FISH workflow alongside your samples. You can use them to identify on slide variability’s in your IHC assays. And interestingly and importantly you can use them to confirm the specificity and sensitivity of your antibody and you’re mentioned probes as we also have FISH reference standards so sorry for the confusion. Most importantly you can use them as a common reference point for routine use in bridging studies or multi centre studies.
A unique feature that we’ve used for some of the labs that are performing this multiplexing , they’ve used these standards, such as the lung panel as multiplexing for doing assay specificity and sensitivity. You could also use these standards for monitoring the accuracy of your quantitative immunofluorescence assay read out and we are also able to customised biomarker specific blocks and slides based on the requirements.
So to conclude, I would like to ask you three important questions. Is your assay optimised? What is the limit of detection of your workflow and are you interested in achieving consistent results.