Benz, S. C. | bioRxiv Science

bioRxiv2021

Deep learning-based risk stratification for HER2-negative breast cancer patients

In this paper, we present our analysis of the tumor microenvironment in digital pathology images to stratify risk in HER2-negative breast cancer patients using clinicopathological, spatial image, and cell-based features in a Coxs proportional hazard model. We start the analysis by processing a set of 304 training pathology images using our in-house pan-cancer trained tumor, stroma, and lymphocyte region identification convolutional neural networks. The next step is computation of spatial regions of interest, namely: lymphocytes within (and adjacent to) tumor, lymphocytes within (and adjacent to) stroma, and stroma within (and adjacent to) tumor areas. Various cell-level information in these regions are then summarized, augmented to clinicopathological data, and linked to patients survival to train a Coxs proportional hazards model. The proposed model outperformed a baseline model based on clinicopathological features only in analysis of an untouched test set of 202 whole slide images with p 8.49E-08 (HR = 0.4273).

pathology↗

bioRxiv2019

Multi-region exome sequencing of ovarian teratomas reveals 2N near-diploid genomes, paucity of somatic mutations, and extensive allelic imbalances shared across mature, immature, and disseminated components

Immature teratoma is a subtype of malignant germ cell tumor of the ovary that occurs most commonly in the first three decades of life, frequently with bilateral ovarian disease. Despite being the second most common malignant germ cell tumor of the ovary, little is known about its genetic underpinnings. Here we performed multi-region whole exome sequencing to interrogate the genetic zygosity, clonal relationship, DNA copy number, and mutational status of 52 pathologically distinct tumor components from 10 females with ovarian immature teratomas, with bilateral tumors present in 5 cases and peritoneal dissemination in 7 cases. We found that ovarian immature teratomas are genetically characterized by 2N near-diploid genomes with extensive loss of heterozygosity and an absence of genes harboring recurrent somatic mutations or known oncogenic variants. All components within a single ovarian tumor (immature teratoma, mature teratoma with different histologic patterns of differentiation, and yolk sac tumor) were found to harbor an identical pattern of loss of heterozygosity across the genome, indicating a shared clonal origin. In contrast, the 4 analyzed bilateral teratomas showed distinct patterns of zygosity changes in the right versus left sided tumors, indicating independent clonal origins. All disseminated teratoma components within the peritoneum (including gliomatosis peritonei) shared a clonal pattern of loss of heterozygosity with either the right or left primary ovarian tumor. The observed genomic loss of heterozygosity patterns indicate that diverse meiotic errors contribute to the formation of ovarian immature teratomas, with 11 out of the 15 genetically distinct clones determined to result from the failure of meiosis I or II. Overall, these findings suggest that copy-neutral loss of heterozygosity resulting from meiotic abnormalities may be sufficient to generate ovarian immature teratomas from germ cells.

cancer biology↗