Search

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., (Copyright © 2024 The Authors. Published by Elsevier Inc. All rights reserved.)

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., (© 2021. The Author(s).)

A high frequency of precursor lesions is a risk factor for cancer in many organ systems but must be precisely quantified. Pelvic serous neoplasia is associated with an estimated increase in frequency of secretory cell outgrowths (SCOUTs) with loss of PAX2 protein (PAX2p) expression (PAX2p-null SCOUTs) in the fallopian tube. However, to confirm this, PAX2p-null SCOUTs must be precisely quantified relative to the epithelial surface. We developed a method by which fallopian tube sections were digitized using an iScan brightfield scanner (BioImagene) and uploaded in Adobe Photoshop CS3 Extended. Pixel length was translated into microns and epithelial length measured with the Magic Wand tool. SCOUTs were expressed as a function of total epithelial perimeter. Frequency, required perimeter length, topographic clustering tendency and effects of age were ascertained. SCOUT frequency per 10 cm was 0-4.60 for cases and 0-1.66 for controls, averaging 0.84 and 0.27, respectively, (P=0.007). Required perimeter length for SCOUT detection was less in serous cancer cases and topographic distribution followed a random pattern without aberrant clustering. Age was also associated with SCOUT frequency (P=0.025) and differences between cancers and controls were still significant after adjusting for age (P=0.001). We describe an efficient method for quantifying epithelial perimeter in the fallopian tube and verify its relevance to precursor frequency. This has important implications for assessing precursor frequency both in the fallopian tube and in other organs-such as prostate, pancreas and colon-where epithelial precursors are integral to carcinogenesis.

Ovarian serous borderline tumors (SBTs) are presumed to originate in the ovarian cortex or peritoneal surface. The pathogenetic role of the fallopian tube (FT) is unclear; however, recently, secretory cell outgrowths (SCOUTs) lacking PAX2 expression were described in benign FTs. This study addressed (1) the differentiation characteristics of SBTs and (2) the frequency of SCOUTs lacking PAX2 expression in the FTs of patients with SBTs and compared (3) SCOUT morphology and (4) PAX2 expression with SBTs. SBTs and FT epithelium shared both ciliated (p73) and secretory (HMFG2) differentiation. PAX2-null SCOUT frequency in FT cross-sections from patients with SBTs was 0.28 (110 of 398) versus 0.112 in benign hysterectomies and nearly 0 in pediatric and postpartum sterilization specimens (P = < 0.001). When adjusted for age, the differences narrowed but remained significant (P = 0.010). SCOUTs were heterogeneous, some displaying ciliated differentiation and papillary architecture. Two cases of discrete multifocal papillary SCOUTs in the FTs were associated with SBTs. All SBTs had heterogeneous PAX2 staining with areas of PAX2 loss. This study shows for the first time that PAX2-null SCOUTs are more common in the oviducts of women with SBTs and that loss of PAX2 expression occurs in most SBTs. These discoveries link both morphologic and functional gene (PAX2) alterations in the oviduct to SBTs, similar to that reported in high-grade serous carcinoma. Further study is warranted to clarify the relationship of the oviduct to serous neoplasia.

Ovarian serous neoplasms can have morphologic overlap with malignant mesothelioma. The distinction is clinically important, yet most studies have failed to identify immunostains that reliably distinguish these 2 tumor types. Recently, transcription factor PAX8 was shown to be a sensitive and relatively specific marker for Müllerian tumors. In addition, some studies suggest that h-caldesmon is sensitive and specific for mesothelioma when compared with serous ovarian tumors. The goal of this study was to evaluate whether PAX8 and h-caldesmon expression can successfully distinguish mesothelioma from serous ovarian tumors. Immunohistochemistry was carried out using PAX8 and h-caldesmon antibodies on archival tissue from 254 ovarian serous tumors and 50 mesothelial tumors. Nuclear and cytoplasmic immunoreactivity were considered positive for PAX8 and h-caldesmon, respectively. PAX8 staining was present in 99% of high-grade serous ovarian carcinomas and all (100%) low-grade ovarian carcinomas and serous borderline tumors; however, only 74% of these cases (188/254) were diffusely positive in more than 50% of tumors cells, and intensity ranged from strong to weak. None of the pleural malignant mesotheliomas were reactive with PAX8. However, 2/23 (9%) peritoneal malignant mesotheliomas showed focal and/or weak staining for PAX8; the remaining cases were negative. Two well-differentiated papillary mesotheliomas and 1 multicystic mesothelioma each showed some staining for PAX8. h-caldesmon was negative in all serous neoplasms and all mesothelial neoplasms, except 1 pleural malignant mesothelioma which showed patchy immunoreactivity. Strong PAX8 staining is highly specific (P<0.00001) for ovarian serous tumors when compared with malignant mesotheliomas of the peritoneum and pleura. The presence of weak staining for PAX8 in the 3 "noninvasive" mesotheliomas questions the use for PAX8 in this differential diagnosis. On the basis of this study, h-caldesmon is not a useful marker for mesothelioma.