Digital pathology and AI offer a host of benefits for pathologists, especially when it comes to breast cancer diagnosis, which is riddled with unique challenges. Partnership and integration are key to tapping into those benefits, and to making the breast cancer diagnostic process better for both pathologists and patients. Earlier this week while at the United States and Canadian Academy of Pathology (USCAP) annual conference in New Orleans, Paige Medical Vice President Dr. Juan Retamero and Mindpeak CEO Felix Faber shared the stage to discuss why integration is important for improving diagnosis, and how incorporating AI into pathology workflows can make for a more direct diagnosis for patients.
Dr. Retamero first explained that “integration” can mean two things: integrating AI tools into the pathology workflow, and more deeply integrating pathology with the other disciplines in the cancer care continuum such as radiology and oncology. As it stands today, pathology is siloed from the other specialties, which could result in decision discrepancies between the physicians that may delay diagnosis and treatment. When pathology labs go digital and their information becomes digitally accessible as opposed to exclusive to the microscope, each stakeholder can easily communicate and make connected decisions, which could reduce misinterpretations and resolve any discrepancies before they reach the patient. Further, many labs currently operate without the support of diagnostic tools like AI, or even without pathology fellows or specialists, which can leave room for human error that could impact patient treatment. AI helps provide pathologists with greater decision-making confidence, efficiency, and reduced subjectivity that could help them avoid misdiagnosis or delays before a patient would be impacted.
Therefore, integration – both of AI into pathology and of pathology into the larger diagnostic chain – is critical. And partnership is what makes integration possible. Paige’s recently announced partnership with Mindpeak was established to put the tools pathologists need for the full spectrum of breast cancer diagnosis in one place. By bringing Mindpeak’s IHC analysis algorithms into the Paige Platform, the partnership makes it easy for labs who are going digital to get up and running with AI-powered workflows quickly, while offering solutions for each of the challenges in breast cancer diagnosis.
To support this, Dr. Retamero highlighted some of the issues at each diagnostic stage and how AI might help, starting with tumor detection. He shared a study that found concordance between pathologists in diagnosing cancer was only 75%, and interestingly, benign cases were overinterpreted 13% of the time.1 Paige Breast*, an AI application for the detection of premalignant and malignant neoplasms, was built with high sensitivity to increase pathologists’ confidence in their breast diagnosis, especially for knowing that those cases the AI calls negative are truly benign. It additionally draws pathologists’ attention to the area on the slide most likely to harbor cancer and highlights cancerous cells while fading out non-tumorous tissue in the TissueMap to help pathologists to focus their attention on what matters and save time during case review.
A similar disconcordance problem exists in lymph node diagnosis; in one study, up to 24% of patients staged by general pathologists later had their N status upstaged when assessed by a specialist.2 Paige Breast Lymph Node*, an AI for identifying lymph node metastases, reduces this subjectivity and has been proven to perform on even small, challenging micromets and ITCs. It also drastically reduces the time lymph node assessment takes, showing in one study to have supported pathologists in delivering their diagnosis 55% faster.3 On the note of efficiency, Dr. Retamero asked the room to raise their hands if they dislike mitotic counting. Unsurprisingly, many pathologists agreed that the process is tedious, time-consuming, and challenging. Paige Breast Mitosis** eases the burden of mitotic counting by detecting the 2 mm2 hotspot on the slide that contains the greatest number of mitoses and proposing an automatic whole-slide score, again working to improve efficiency and confidence.
Mindpeak CEO Felix Faber then walked through the next step in the diagnostic process, IHC quantification. This, he said, is another area where there can be massive discrepancies across individual pathologists, institutions, and even geographical regions. Mindpeak’s AI algorithms were designed to be agnostic to variations in stainers and scanners to reduce this subjectivity between regions or sites. Their algorithms for analyzing breast cancer biomarkers, including Ki-67, ER & PR, and HER2, each identify and classify tumor cells, providing automatic scores for regions of interest or whole slides based on current biomarker guidelines. This has been proven not only to reduce the time it takes to calculate a score by over 80%4, but to enhance diagnostic accuracy, which is especially important for biomarker review as this has direct impacts on patient treatment.
Through the Paige and Mindpeak partnership, 5 total applications, including these breast tools, will be available on the Paige Platform. Having all of these tools together means pathologists can move through their workflow more efficiently, without learning more than one system or logging in and out of multiple platforms. This makes it simple for labs to adopt digital pathology and tap into its many benefits, and most importantly, sets them up for success as AI continues to evolve in cancer care.
While these tools will advance the future of pathology, AI will never be a replacement for pathologists. Rather, it amplifies the unique set of skills that pathologists already possess, allowing them to complete their diagnosis faster and feel more confident in their decision. In this way, the use of AI supports patients in receiving a more direct diagnosis, which we hope creates a world where the right patients get the right treatment at the right time.
Learn more about Paige & Mindpeak’s integration or trial it free.
References
1Elmore JG, Longton GM, Carney PA, et al. Diagnostic Concordance Among Pathologists Interpreting Breast Biopsy Specimens. JAMA. 2015;313(11):1122–1132. doi:10.1001/jama.2015.1405
2Vestjens JHMJ, Pepels MJ, de Boer M, et al. Relevant impact of central pathology review on nodal classification in individual breast cancer patients. Ann Oncol. 2012;23(10):2561-2566. doi:10.1093/annonc/mds072
3Based on an investigational clinical study involving 3 pathologists and data from 148 patients.
4Colón E et al. (2023): AI in Combination with the Global Counting Methodology Recommended by the International Ki67 in Breast Cancer Working Group Identifies More Patients Eligible for Treatment and Increases Speed by 8-fold.
*In European Union & United Kingdom, mentioned Paige products are CE-IVD and UKCA marked for clinical use. In the United States and where research use is permitted, Paige Breast, Paige Breast Lymph Node, and HER2Complete are limited to Research Use Only and not for use in diagnostic procedures.
**Paige Breast Mitosis is currently released for evaluation/demo purposes only and not for use in diagnostic procedures.
