Biology subjects

Braunstein, L. Z.

Publications and source records attributed to Braunstein, L. Z..

3 recordsLinked to original sources

bioRxiv2025

Integrated histopathologic modeling of detailed tumor subtypes and actionable biomarkers

Accurate cancer subtyping with accompanying molecular characterization is critical for precision oncology. While machine learning approaches have been applied to both digital pathology and cancer genomics, previous work has been limited in sample size and has typically aggregated granular cancer subtypes into coarse groupings, likely obfuscating informative molecular and prognostic associations and phenotypic variation of more detailed tumor subtypes. Accordingly, we collated 378,123 hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) with matched targeted DNA clinical sequencing results and OncoTree detailed cancer subtypes from a real-world cohort of 71,142 patients. Using this scaled, granular dataset and a cancer subtype knowledge graph, we developed Mosaic: a family of calibrated machine learning models using H&E WSI embeddings to classify tumors and identify molecular phenotypes across 163 detailed subtypes. The cancer subtyping module (Aeon) achieved an area under the receiver operating characteristic curve (AUROC) of 0.992 overall, with 161/163 subtypes reaching an AUROC [≥] 0.90 and improved performance over a state-of-the-art genomics-based classifier. The genomic inference module (Paladin) achieved an AUROC [≥] 0.80 for 167 pairs of detailed subtypes and genomic targets. We further used the learned histopathologic representations to i) identify key associations of the histopathologic embeddings with clinical biomarkers; ii) identify unsupervised sub-clusters of tumors with genomic determinants of tumor phenotype; iii) specify granular diagnoses for cancers of unknown primary, evaluated by genomic associations and expected clinical outcome distributions; iv) annotate functional significance for variants of uncertain significance (VUS); and v) identify cases that mimic the phenotypic effect of known DNA variants on H&E in the absence of detectable DNA alterations. Taken together, this work advances our understanding of phenotypic variation of granular tumor subtypes, their relevance to enhanced diagnostics, and their potential utility in risk stratification with multimodal machine learning in cancer.

cancer biology↗

bioRxiv2020

Age-related expression of SARS-CoV-2 priming protease TMPRSS2 in the developing lung

The SARS-CoV-2 novel coronavirus global pandemic (COVID-19) has led to millions of cases and hundreds of thousands of deaths around the globe. While the elderly appear at high risk for severe disease, hospitalizations and deaths due to SARS-CoV-2 among children have been relatively rare. Integrating single-cell RNA sequencing (scRNA-seq) of the developing mouse lung with temporally-resolved RNA-in-situ hybridization (ISH) in mouse and human lung tissue, we found that expression of SARS-CoV-2 Spike protein primer TMPRSS2 was highest in ciliated cells and type I alveolar epithelial cells (AT1), and TMPRSS2 expression was increased with aging in mice and humans. Analysis of autopsy tissue from fatal COVID-19 cases revealed SARS-CoV-2 RNA was detected most frequently in ciliated and secretory cells in the airway epithelium and AT1 cells in the peripheral lung. SARS-CoV-2 RNA was highly colocalized in cells expressing TMPRSS2. Together, these data demonstrate the cellular spectrum infected by SARS-CoV-2 in the lung epithelium, and suggest that developmental regulation of TMPRSS2 may underlie the relative protection of infants and children from severe respiratory illness.

developmental biology↗

bioRxiv2019

Tumor mutational landscape is a record of the pre-malignant state

Chromatin structure has a major influence on the cell-specific density of somatic mutations along the cancer genome. Here, we present a pan-cancer study in which we searched for the putative cancer cell-of-origin of 2,550 whole genomes, representing 32 cancer types by matching their mutational landscape to the regional patterns of chromatin modifications ascertained in 104 normal tissue types. We found that, in almost all cancer types, the cell-of-origin can be predicted solely from their DNA sequences. Our analysis validated the hypothesis that high-grade serous ovarian cancer originates in the fallopian tube and identified distinct origins of breast cancer subtypes. We also demonstrated that the technique is equally capable of identifying the cell-of-origin for a series of 2,044 metastatic samples from 22 of the tumor types available as primaries. Moreover, cancer drivers, whether inherited or acquired, reside in active chromatin regions in the respective cell-of-origin. Taken together, our findings highlight that many somatic mutations accumulate while the chromatin structure of the cell-of-origin is maintained and that this historical record, captured in the DNA, can be used to identify the often elusive cancer cell-of-origin.

genomics↗