Poster Spotlight Session 3: Insights from Single Cell, Spatial, and Artificial Intelligence Approaches
Tuesday, December 5 • 5:30 p.m. – 6:30 p.m. • Stars at Night Ballroom 1-2
Presentation: Tumor immune microenvironment modulates resistance to estrogen suppression in ER+ breast cancer
Fabiana Napolitano, MD
UT Southwestern Medical Center, Simmons Comprehensive Cancer Center,
Dallas, Texas
What is your presentation about?
In my presentation, I will discuss how the application of spatial transcriptomic was able to uncover the composition of the tumor immune microenvironment (TIME) in estrogen receptor-positive (ER+) breast cancer (BC) resistant to endocrine treatment. The major therapeutic strategy for patients with ER+ BC is estrogen suppression, but approximately 20-25% of patients diagnosed with ER+ early BC show intrinsic resistance to endocrine treatment, and thus present a higher risk of relapse. Moreover, in the past few years, increasing evidence has demonstrated how the TIME plays an important role in the tumorigenesis, progression, and metastatic spread of BC.
Therefore, we used a short-term presurgical treatment with endocrine therapy to categorize patients’ sensitivity to estrogen suppression and investigated the association between resistance to endocrine therapy and TIME composition, in particular in terms of immune cell subtypes and pathway regulation.
What makes this topic important in 2023?
Currently, the primary cause of death in patients with breast cancer is advanced disease. Therefore, it is of utmost importance to improve prognostic and predictive categorization, particularly to correctly select the subgroup of patients whose tumors display aggressive features and may require treatment escalation. Moreover, according to the results of the I-SPY2 trial, a group of patients with ER+ BC may benefit from the addition of immune checkpoint inhibitors. Our findings may be applied to identify tumors with aggressive biological and molecular features that are resistant to endocrine therapy but show an exploitable TIME. Ultimately, these results will help identify patients with worse prognosis and endocrine resistance, which may be sensitive to novel combination therapies.
How did you get involved in this particular area of breast cancer research, care, or advocacy?
I am interested in understanding how tumors and their microenvironment shape each other through continuous interactions. Moreover, the systemic treatments that patients undergo are a third factor in this equation, inducing changes in both the tumor and the TIME. I believe that being able to understand how all these components interact will help physicians better define the therapeutic strategy for patients with breast cancer.
Presentation: Clinical implementation of artificial-intelligence-assisted detection of breast cancer metastases in sentinel lymph nodes: saving costs and time (the CONFIDENT-B trial)
Carmen van Dooijeweert, MD, PhD
University Medical Centre Utrecht,
Utrecht, Netherlands
What is your presentation about?
My presentation will focus on the implementation and impact of an artificial intelligence (AI)-assisted clinical workflow in the detection of sentinel lymph node (SN) metastases in breast cancer patients. Metastases in these SNs are a critical factor affecting patient survival and guiding treatment decisions. However, the current diagnostic process involves the (tedious) assessment of standard hematoxylin-eosin (HE) stained slides, and if metastases are not detected upon first assessment, additional costly immunohistochemistry (IHC) is performed to ensure no metastases are missed. In this prospective and pragmatic trial (CONFIDENT-B) in the University Medical Center Utrecht (the Netherlands) we aimed to determine to what extent the use of AI can reduce the need for costly IHC, while maintaining diagnostic safety standards. We enrolled 190 unilateral SN-specimens from 182 patients with breast cancer between September 2022 and May 2023. These specimens were divided bi-weekly into two arms: a control arm and an intervention arm. In the control arm, an expert breast pathologist assessed the HE-stained sections traditionally, while in the intervention arm, an expert breast pathologist used the ‘Metastasis-Detection-App’ (Visiopharm) to assist their assessment. In both arms, additional IHC was performed if no metastases were identified. The results were promising. AI assistance was safe and significantly reduced the risk of IHC-use per detected case of SN-metastases. This not only saved costs but also reduced assessment time for pathologists, who stated that using AI made their work more enjoyable.
What makes this topic important in 2023?
The topic of AI-implementation is important in 2023 for multiple reasons. First, because for over a decade many studies have been published on “promising” algorithms. Yet, prospective studies on clinical implementation are lacking. With our trial, to the best of our knowledge, we are the first to do so in (digital) pathology. Second, we show that implementation of an AI-assisted workflow leads to a significant reduction of IHC-use and subsequent costs for the detection of SN-metastases in breast cancer patients, while saving pathologists time and making their work more enjoyable. Importantly, AI-implementation during this trial was safe and patients were not at risk of an inferior diagnosis. These cost savings are highly relevant in the current era of skyrocketing health care costs and limited resources (i.e. worldwide shortage of pathologists). As opposed to many innovations, the implementation of AI in pathology laboratories that work fully digital may reduce health care costs and shows important benefits for pathologists and the laboratory workflow. Such tangible cost- and time-savings demonstrate the potential of AI-implementation to keep accurate diagnostic pathology affordable and accessible.
How did you get involved in this particular area of breast cancer research, care, or advocacy?
After obtaining my medical degree I started my PhD on variation in breast cancer biomarker assessment (receptor assessment and histologic grading) by pathologists. In my research we observed significant differences in biomarker assessment between pathology laboratories and between pathologists within these laboratories. Although we showed that improvement is possible by feedback-reports and e-learning, (significant) variation remained. As the pathology department of the UMC Utrecht is one of the frontrunners in digital pathology, it made sense to investigate whether artificial intelligence could assist human intelligence in many of their tasks, like for example biomarker assessment. For myself, this research project was part of my postgraduate master in (clinical) epidemiology, which I recently obtained. Our group of AI-(implementation)-researchers is highly motivated to implement many more algorithms in clinical practice, and investigate its potential, in a safe manner.
Presentation: Morphometric signature identifies ductal carcinoma in situ of the breast with low risk of progression to invasive breast cancer
Marcelo Sobral-Leite, PharmD, PhD
Netherlands Cancer Institute,
Amsterdam, Netherlands
What is your presentation about?
Ductal carcinoma in situ (DCIS) is a potential precursor of invasive breast cancer (IBC). However, most DCIS lesions do not progress to invasive breast cancer. When it is not possible to distinguish progressive from non-progressive DCIS, almost all women are intensively treated. It exposes many women to undesirable overtreatment.
We aimed to find clinical biomarkers able to distinguish non-progressive from progressive DCIS. For that, we developed an AI-based DCIS morphometric analysis (AIDmap) to detect DCIS pathology and measure DCIS morphological structures in H&E slides. This AI-tool showed potential to detect which DCIS would not progress and therefore might not need treatment like radiotherapy.
What makes this topic important in 2023?
Developing artificial intelligence (AI) tools in breast cancer digital pathology holds significant importance due to several reasons:
Accuracy and Efficiency: AI-powered tools can analyze vast amounts of pathology data, such as digitized images of breast tissue samples, with high accuracy and speed. These tools have the potential to assist pathologists in accurately detecting and classifying breast cancer cells and abnormalities more efficiently than traditional manual methods.
Early Detection and Diagnosis: AI can aid in early detection by identifying subtle patterns or features in pathology images that might not be easily recognizable to the human eye. Early detection is crucial in improving treatment outcomes and increasing survival rates for breast cancer patients.
Personalized Treatment: AI-driven analysis can help pathologists assess the aggressiveness of tumors and predict patient outcomes based on the tumor’s characteristics. This information contributes to personalized treatment plans, ensuring patients receive the most suitable and effective therapies.
How did you get involved in this particular area of breast cancer research, care, or advocacy?
I have worked in breast cancer research for a long period of time. I have training in oncology and clinical pharmacy but I have focused my work on the research. In my PhD I have searched for biomarkers of breast cancer outcomes. During my postdoc I saw the potential to use AI-tools for the benefit of patients.