Visiopharm Precision Pathology

Pathology deserves standardization, patients expect it

Pathology needs standardization

Staining quality & sufficiency have a significant impact on the interpretive accuracy of the diagnostic tests based on IHC [1]. The human costs of inaccurate diagnostic interpretation can be high, and so are the financial costs.

AI and image analysis can play an important role and can be used for characterizing and quantifying staining quality. Such technology can be used both as decision support for organizations, assisting them in achieving both standardization and scalability and ultimately ensuring that the correct decisions are taken.

The quality challenge

Tissue biomarkers offer a diagnostic window into the tumor micro-environment, allowing scientists to discover new predictive biomarkers, for patient stratification and treatment selection.
However, the journey from biopsy to diagnosis has several steps, each with the potential to contribute to errors. 

Biopsy and fixation

Due to errors in fixation, up to 20% of biopsies are unfit for diagnosis. This leads to a longer turn-around processing time, and often this requires a new sample to be collected [2].

Stain quality

20-30% of labs have insufficient stain quality for diagnosis. This leads to a delayed and incorrect diagnosis. Read more about it.

Image quality

Errors in digitization lead to poor image quality that can result in images with dark nuclei, images that are out of focus, or images that miss critical parts of the tissue. This can lead to longer turnaround times, unacceptable and borderline acceptable stain quality scores, and a delay in diagnosis [3].

Diagnostic evaluation

Inter- and intra-observer variability during interpretation and assessment of complex biomarkers may lead to sub-optimal and incorrect patient treatment.
Vimeo

By loading the video, you agree to Vimeo's privacy policy.
Learn more

Load video

Implications of stain quality in the diagnostic lab

Dr. Mogens Vyberg provides examples from his time at NordicQC to discuss the need to standardize sample staining in the clinic.

Unmet needs in stain quality

In quality runs conducted by NordiQC from 2003-2015,  about 30% of the stains in the general module and 20% in the Breast Cancer Module showed insufficient staining as assessed by the external quality assessment (EQA). While EQA’s give laboratories valuable feedback on the performance of laboratories regularly, there are typically at least 3 to 6 months between assessments due to bandwidth limitations at EQAs.

The variability between EQA runs represents a real problem in terms of standardization and a concomitant lack of robustness in tissue diagnostic testing.  This is a problem both in routine diagnostics and in clinical trials.

Fig. Two sections from the same tissue sample, stained with PD-L1 at two different labs. Lab A has insufficient staining, while Lab B has sufficient staining.

Join the early
access program

Be at the forefront of biomarker quality & standardization.

We are currently in the development of Quality solutions; there is an opportunity for you to participate. Join our Qualitopix early access program to participate in surveys, focus groups, prototype feedback or as an early test site.

Begin your journey towards standardization by implementing AI

In a proof-of-concept project, we wanted to show the potential of AI and image analysis in combination with controls. As part of a pilot for the assessment of PD-L1 for triple-negative breast cancer (TNBC), an EQA sent out unstained tissue slides to 16 participating pathology labs.
These tissue slides contained tonsil, breast cancer tissues (low and high expression), and four cell lines from HistoCyte. After assessment by the pathologist expert panel, the slides were digitized and analyzed with AI and image analysis.
Tonsil
Triple-negative breast cancer
Cell line (5%-15%)
Cell lines (75%)
Tonsil

Tonsil without and with AI-image analysis results overlaid.

Triple-negative breast cancer

TNBC without and with AI-image analysis results overlaid.

Cell line (5%-15%)

Cell line (5%-15%) without and with AI-image analysis results overlaid.

Cell lines (75%)

Cell lines (75%) without and with AI-image analysis results overlaid.

The results from the pilot suggest that there is predictive power in the use of AI for the analysis of standard controls (tonsil and TNBC tissues) and analysis of standardized cell lines. See our poster on the pilot here.
Such technology can be used as decision support for EQA organizations, to achieve standardization and scalability, and offer stain quality control for pathology labs. AI tools will allow organizations to see trends and fluctuations in quality, and adjust protocols accordingly.
Longer-term, this technology may provide 100% on-slide quality control, introducing quantitative statistical techniques commonly used in the clinical pathology lab, to the histology lab.

Latest news

What are your challenges in standardizing your assessment of biomarkers?
Share with us!

Thank you for sending us a message
We have a passion for quality and are committed to providing end-to-end quality solutions for tissue biomarkers. We are taking radical and innovative approaches towards the standardized assessment of biomarkers.

References

[1] Vyberg M, Nielsen S. Proficiency testing in immunohistochemistry – experiences from Nordic Immunohistochemical Quality Control (NordiQC). Virchows Arch. 2016 Jan;468(1):19-29. DOI: 10.1007/s00428-015-1829-1. Epub 2015 Aug 26. PMID: 26306713; PMCID: PMC4751198.

[2] Sophia Apple, MD, MS, et al., The Effect of Delay in Fixation, Different Fixatives, and Duration of Fixation in Estrogen and Progesterone Receptor Results in Breast Carcinoma, American Journal of Clinical Pathology, Volume 135, Issue 4, April 2011, Pages 592–598.

[3] Kohlberger T, Liu Y, Moran M, et al. Whole-slide image focus quality: Automatic assessment and impact on ai cancer detection. J Pathol Inform. 2019;10(1):39. doi:10.4103/jpi.jpi_11_19

Success
Your message has been successfully sent!