Opening the Black Box: Spatial Transcriptomics and the Relevance of Artificial Intelligence-Detected Prognostic Regions in High-Grade Serous Carcinoma.

Image-based deep learning models are used to extract new information from standard hematoxylin and eosin pathology slides; however, biological interpretation of the features detected by artificial intelligence (AI) remains a challenge. High-grade serous carcinoma of the ovary (HGSC) is characterized by aggressive behavior and chemotherapy resistance, but also exhibits striking variability in outcome. Our understanding of this disease is limited, partly due to considerable tumor heterogeneity. We previously trained an AI model to identify HGSC tumor regions that are highly associated with outcome status but are indistinguishable by conventional morphologic methods. Here, we applied spatially resolved transcriptomics to further profile the AI-identified tumor regions in 16 patients (8 per outcome group) and identify molecular features related to disease outcome in patients who underwent primary debulking surgery and platinum-based chemotherapy. We examined formalin-fixed paraffin-embedded tissue from (1) regions identified by the AI model as highly associated with short or extended chemotherapy response, and (2) background tumor regions (not identified by the AI model as highly associated with outcome status) from the same tumors. We show that the transcriptomic profiles of AI-identified regions are more distinct than background regions from the same tumors, are superior in predicting outcome, and differ in several pathways including those associated with chemoresistance in HGSC. Further, we find that poor outcome and good outcome regions are enriched by different tumor subpopulations, suggesting distinctive interaction patterns. In summary, our work presents proof of concept that AI-guided spatial transcriptomic analysis improves recognition of biologic features relevant to patient outcomes.

Artificial intelligence-based image analysis can predict outcome in high-grade serous carcinoma via histology alone.

High-grade extrauterine serous carcinoma (HGSC) is an aggressive tumor with high rates of recurrence, frequent chemotherapy resistance, and overall 5-year survival of less than 50%. Beyond determining and confirming the diagnosis itself, pathologist review of histologic slides provides no prognostic or predictive information, which is in sharp contrast to almost all other carcinoma types. Deep-learning based image analysis has recently been able to predict outcome and/or identify morphology-based representations of underlying molecular alterations in other tumor types, such as colorectal carcinoma, lung carcinoma, breast carcinoma, and melanoma. Using a carefully stratified HGSC patient cohort consisting of women (n = 30) with similar presentations who experienced very different treatment responses (platinum free intervals of either ≤ 6 months or ≥ 18 months), we used whole slide images (WSI, n = 205) to train a convolutional neural network. The neural network was trained, in three steps, to identify morphologic regions (digital biomarkers) that are highly associating with one or the other treatment response group. We tested the classifier using a separate 22 slide test set, and 18/22 slides were correctly classified. We show that a neural network based approach can discriminate extremes in patient response to primary platinum-based chemotherapy with high sensitivity (73%) and specificity (91%). These proof-of-concept results are novel, because for the first time, prospective prognostic information is identified specifically within HGSC tumor morphology.

Real-world features associated with cancer-related venous thromboembolic events.

BACKGROUND: The incidence of venous thromboembolism (VTE) is 1-2/1000 individuals. Patients with cancer, especially during chemotherapy, are at enhanced risk, but real-world data on factors associated with VTE events are still scarce. AIM: The aim of this retrospective study was to survey the incidence of VTE based on a large hospital database, and to identify comorbidities and features associated with VTE events. We focused on cancer-related VTE events and on factors indicating increased VTE risk during chemotherapy. METHODS: The cohort included patients treated at Turku University Hospital during years 2005-2013. Health information was derived and analysed from multiple electronic databases. The diagnoses of VTE and all comorbidities, including type of cancer, were based on International Classification of Diseases 10th Revision coding. For further analysis, we focused on 16 common types of cancers treated with chemotherapy. Age, gender, surgery, radiotherapy, distant metastasis, available laboratory values and platinum-based chemotherapy were evaluated for VTE group, and associations were estimated by Cox regression analyses. RESULTS: The entire database contained information from 495 089 patients, of whom 5452 (1.1%) had a VTE diagnosis. Among individuals with VTE, 1437 (26.4%) had diagnosis of coronary heart disease and 1467 (26.9%) had cancer diagnosis. Among 7778 patients with cancer treated with chemotherapy, 282 (3.6%) had a VTE, platinum-based chemotherapy being a major risk factor (HR 1.77, 95% CI 1.40 to 2.24, p<0.001). In multivariate analysis, elevated blood neutrophil counts (>3.25×109 cells/L, HR 1.96, 95% CI 1.33 to 2.89, p<0.001) and plasma creatinine (>62.5 µmol/L; HR 1.60, 95% CI 1.21 to 2.13, p=0.001) values were independent indicators of increased VTE risk during chemotherapy. CONCLUSIONS: Longitudinal electronic health record analysis provides a powerful tool to gather meaningful real-world information to study clinical associations, like comorbidities, and to identify markers associated with VTE. The combination of various clinical and laboratory variables could be used for VTE risk evaluation and targeted prevention.