MorphoITH: A Framework for Deconvolving Intra-Tumor Heterogeneity Using Tissue Morphology
Authors:
Aleksandra Weronika Nielsen,
Hafez Eslami Manoochehri,
Hua Zhong,
Vandana Panwar,
Vipul Jarmale,
Jay Jasti,
Mehrdad Nourani,
Dinesh Rakheja,
James Brugarolas,
Payal Kapur,
Satwik Rajaram
Abstract:
The ability of tumors to evolve and adapt by developing subclones in different genetic and epigenetic states is a major challenge in oncology. Traditional tools like multi-regional sequencing used to study tumor evolution and the resultant intra-tumor heterogeneity (ITH) are often impractical because of their resource-intensiveness and limited scalability. Here, we present MorphoITH, a novel frame…
▽ More
The ability of tumors to evolve and adapt by developing subclones in different genetic and epigenetic states is a major challenge in oncology. Traditional tools like multi-regional sequencing used to study tumor evolution and the resultant intra-tumor heterogeneity (ITH) are often impractical because of their resource-intensiveness and limited scalability. Here, we present MorphoITH, a novel framework that leverages histopathology slides to deconvolve molecular ITH through tissue morphology. MorphoITH integrates a self-supervised deep learning similarity measure to capture phenotypic variation across multiple dimensions (cytology, architecture, and microenvironment) with rigorous methods to eliminate spurious sources of variation. Using a prototype of ITH, clear cell renal cell carcinoma (ccRCC), we show that MorphoITH captures clinically-significant biological features, such as vascular architecture and nuclear grades. Furthermore, we find that MorphoITH recognizes differential biological states corresponding to subclonal changes in key driver genes (BAP1/PBRM1/SETD2). Finally, by applying MorphoITH to a multi-regional sequencing experiment, we postulate evolutionary trajectories that largely recapitulate genetic evolution. In summary, MorphoITH provides a scalable phenotypic lens that bridges the gap between histopathology and genomics, advancing precision oncology.
△ Less
Submitted 2 February, 2025;
originally announced February 2025.
Histopathology Based AI Model Predicts Anti-Angiogenic Therapy Response in Renal Cancer Clinical Trial
Authors:
Jay Jasti,
Hua Zhong,
Vandana Panwar,
Vipul Jarmale,
Jeffrey Miyata,
Deyssy Carrillo,
Alana Christie,
Dinesh Rakheja,
Zora Modrusan,
Edward Ernest Kadel III,
Niha Beig,
Mahrukh Huseni,
James Brugarolas,
Payal Kapur,
Satwik Rajaram
Abstract:
Predictive biomarkers of treatment response are lacking for metastatic clear cell renal cell carcinoma (ccRCC), a tumor type that is treated with angiogenesis inhibitors, immune checkpoint inhibitors, mTOR inhibitors and a HIF2 inhibitor. The Angioscore, an RNA-based quantification of angiogenesis, is arguably the best candidate to predict anti-angiogenic (AA) response. However, the clinical adopt…
▽ More
Predictive biomarkers of treatment response are lacking for metastatic clear cell renal cell carcinoma (ccRCC), a tumor type that is treated with angiogenesis inhibitors, immune checkpoint inhibitors, mTOR inhibitors and a HIF2 inhibitor. The Angioscore, an RNA-based quantification of angiogenesis, is arguably the best candidate to predict anti-angiogenic (AA) response. However, the clinical adoption of transcriptomic assays faces several challenges including standardization, time delay, and high cost. Further, ccRCC tumors are highly heterogenous, and sampling multiple areas for sequencing is impractical. Here we present a novel deep learning (DL) approach to predict the Angioscore from ubiquitous histopathology slides. To overcome the lack of interpretability, one of the biggest limitations of typical DL models, our model produces a visual vascular network which is the basis of the model's prediction. To test its reliability, we applied this model to multiple cohorts including a clinical trial dataset. Our model accurately predicts the RNA-based Angioscore on multiple independent cohorts (spearman correlations of 0.77 and 0.73). Further, the predictions help unravel meaningful biology such as association of angiogenesis with grade, stage, and driver mutation status. Finally, we find our model can predict response to AA therapy, in both a real-world cohort and the IMmotion150 clinical trial. The predictive power of our model vastly exceeds that of CD31, a marker of vasculature, and nearly rivals the performance (c-index 0.66 vs 0.67) of the ground truth RNA-based Angioscore at a fraction of the cost. By providing a robust yet interpretable prediction of the Angioscore from histopathology slides alone, our approach offers insights into angiogenesis biology and AA treatment response.
△ Less
Submitted 28 May, 2024;
originally announced May 2024.
