In a recent publication, the inadequate calibration (and use) of the primary antibody is one of the most common causes cited of a sub-optimal stain, with approximately 90% of the ‘insufficient’ results were characterized by either a signal that was too weak or a false negative staining. These results comes from a retrospective study, spanning 2003-2015 and looking at over 26,000 slides. The NordiQC 2015 publication classifies 29% as ‘insufficient’, 30% as ‘good’ and 41% as ‘optimal’.
Detailing out the most common causes for insufficient results, listed by NordiQC as:
- Inadequate calibration of the primary antibody,
- Use of an inadequate primary antibody,
- Inappropriate choice of epitope retrieval method,
- Insufficient heat induced epitope retrieval (HIER) and;
- Use of an inadequate detection kit
So what does this mean for patients? In a recent study, Kaufman et al. show that, out of 552 patients who were tested for HER2 and came out negative, 22 were actually positive, and so likely did not receive potentially efficacious therapy because their tumor HER2-positivity was not accurately identified. To put that into a (slightly irrelevant) comparison, you’re 160x more likely to be diagnosed wrongfully if you’re having a HER2 IHC test than get struck by lightning in your entire lifetime.
Regardless of the accuracy, the importance of immunohistochemistry (IHC) to pathology is undeniable and here to stay. Since Albert Coons and colleagues first described IHC in 1941, using antibodies to visualise tissues infected with pneumococcal antigens, the technique has become indispensable and is one of the most-used diagnostic tests worldwide. It is also an important factor in many therapeutic decisions, for example deciding whether or not to treat breast cancer with trastuzumab (Herceptin), which is only effective in those cancers that over express human epidermal growth factor-2 (HER2).
A large challenge is that clinical samples generally contain varying levels of the target antigen, and those with low levels of the antigen can lead to weak staining and false negative results. An example of the impact of misdiagnosis was an EQA study on breast cancer. The authors found that less than half (48%) the diagnoses of atypical hyperplasia were concordant between the expert panel and pathologists. Worryingly, over a third (35%) of the cases were under-interpreted, meaning such women would not receive sufficient medical treatment while 17% of the cases were over-interpreted, resulting in such women potentially undergoing unnecessary invasive therapy.
This study is far from unique and many other studies and publications have shown similar findings. The table below demonstrates several more examples looking in this case at HER2:
|Error Rate||Number of Labs||Source|
|4%||1||Kaufman A P et al; Cancer Vol 120, Issue 17, 2014|
|<3%||18||Jefferson T et al; Appl Immuno Mol Morphol Vol 19, N 05, 2009|
|18.3%||109||Bogen et al; Arch Pathol Lab Med Vol 132, Feb 2008|
|Up to 20%||1||Wolff AC et al; Journal of Clinical Oncology 25:118-145 2008|
|38%||93||Rhodes A et al; Am J Clin Pathol, vol 118, 408-417 2002|
Comparison of LDTs and FDA/CE-IVD Assays
This statistic gets even more pronounced when you look at laboratory developed tests (aka LDTs). 96% (113/118) labs that use a closed-system when performing HER2 testing (with protocol-standardised FDA approved kits available) achieve an optimal sufficiency scoring – the most used kit in the study was PATHWAY® rmAb clone 4B5, from Ventana (118 of 333 labs used this kit). However, in contrast, as all Ki67 tests are currently LDTs with no FDA-test available, a large inter-laboratory variability can be seen in staining patterns. Standardizing and optimizing protocols for LDTs is particularly challenging since, they require a large number of tissue samples (>100) that are known to express the epitope of interest at levels found in diseased tissue and, considerable effort and time are needed to prepare these control samples, calibrate and optimize the protocol.
Figure – Run 19 of the HER2 EQA scheme. These results strongly suggest that all labs could use better optimization and standardization, with LDTs needing special attention.
IHC performed on the same type of human tissue (for example breast biopsies) can vary considerably between laboratories due to the way in which each laboratory performs each stage of the process (pre-analytical, analytical and post-analytical) , with inaccurate or false-negative results potentially leading to incorrect therapeutic decisions, impacting the efficiency of operations of the pathology lab, and potentially having a significant impact on patient care.
Tissue samples for IHC are heterogeneous since they come from a wide range of genetic backgrounds and are obtained at different stages of disease progression (thus having different levels and patterns of epitope expression). This, added to the differences in sample collection, fixing, FFPE tissue slice thickness, antigen retrieval method, antibody type and titre, lead to large variations in staining patterns.
“HER2 is overexpressed in 36 out of 100,000 women”
Trastuzumab, the treatment for HER2 positive breast cancers, is unfortunately ineffective in other breast cancer types, hence, it is important to identify the type of breast cancer in order to get the correct therapy. The accuracy of HER2 is therefore critical to give a patient the best chance of survival. Furthermore, given breast cancer incidence rates of 89.7 (Western Europe) and 122 (US) per 100,000 women, this means that HER2 is potentially overexpressed in 27 (Western Europe) to 36 (US) women per 100,000. So highly prevalent.
There are many companies and societies seeking to rectify this situation and it is good to see companies, like Abcam and Horizon Discovery, who are committed to raising antibody standards (one of the main causes of incorrect results). Other societies and EQA schemes such as cIQC, NordiQC and IQNpath are also constructively creating guidelines to help the industry and technological advancements has meant that companies such as Visiopharm have also been able to quantify the quality of IHC staining through Quantitative Digital Pathology (QDP).
There is still a lot more work to be done and, I believe, that a 100% accurate test is achievable – however, it will require industry collaboration at all levels. For example, by raising antibody standards (as one of the highest contributing factors towards an insufficient IHC result), up to 9 of every 10 current sub-optimal IHC test results could potentially be avoided. Something that I can already see happening and I look forward to seeing this trend continue…