Dataset Information

Deep Morphology Learning Enhances Ex Vivo Drug Profiling-Based Precision Medicine.

ABSTRACT: Drug testing in patient biopsy-derived cells can identify potent treatments for patients suffering from relapsed or refractory hematologic cancers. Here we investigate the use of weakly supervised deep learning on cell morphologies (DML) to complement diagnostic marker-based identification of malignant and nonmalignant cells in drug testing. Across 390 biopsies from 289 patients with diverse blood cancers, DML-based drug responses show improved reproducibility and clustering of drugs with the same mode of action. DML does so by adapting to batch effects and by autonomously recognizing disease-associated cell morphologies. In a post hoc analysis of 66 patients, DML-recommended treatments led to improved progression-free survival compared with marker-based recommendations and physician's choice-based treatments. Treatments recommended by both immunofluorescence and DML doubled the fraction of patients achieving exceptional clinical responses. Thus, DML-enhanced ex vivo drug screening is a promising tool in the identification of effective personalized treatments.

Significance

We have recently demonstrated that image-based drug screening in patient samples identifies effective treatment options for patients with advanced blood cancers. Here we show that using deep learning to identify malignant and nonmalignant cells by morphology improves such screens. The presented workflow is robust, automatable, and compatible with clinical routine. This article is highlighted in the In This Issue feature, p. 476.

SUBMITTER: Heinemann T

PROVIDER: S-EPMC9894727 | biostudies-literature | 2022 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep Morphology Learning Enhances Ex Vivo Drug Profiling-Based Precision Medicine.

Heinemann Tim T Kornauth Christoph C Severin Yannik Y Vladimer Gregory I GI Pemovska Tea T Hadzijusufovic Emir E Agis Hermine H Krauth Maria-Theresa MT Sperr Wolfgang R WR Valent Peter P Jäger Ulrich U Simonitsch-Klupp Ingrid I Superti-Furga Giulio G Staber Philipp B PB Snijder Berend B

Blood cancer discovery 20221101 6

Drug testing in patient biopsy-derived cells can identify potent treatments for patients suffering from relapsed or refractory hematologic cancers. Here we investigate the use of weakly supervised deep learning on cell morphologies (DML) to complement diagnostic marker-based identification of malignant and nonmalignant cells in drug testing. Across 390 biopsies from 289 patients with diverse blood cancers, DML-based drug responses show improved reproducibility and clustering of drugs with the sa ...[more]

PMID: 36125297

Similar Datasets

Project description:BACKGROUND: Understanding individual patient host response to viruses is key to designing optimal personalized therapy. Unsurprisingly, in-vivo human experimentation to understand individualized dynamic response of the transcriptome to viruses are rarely studied because of the obvious limitations stemming from ethical considerations of the clinical risk. OBJECTIVE: In this rhinovirus study, we first hypothesized that ex-vivo human cells response to virus can serve as a proxy for otherwise controversial in-vivo human experimentation. We further hypothesized that the N-of-1-pathways framework, previously validated in cancer, can be effective in understanding the more subtle individual transcriptomic response to viral infection. METHOD: N-of-1-pathways computes a significance score for a given list of gene sets at the patient level, using merely the 'omics profiles of two paired samples as input. We extracted the peripheral blood mononuclear cells (PBMC) of four human subjects, aliquoted in two paired samples, one subjected to ex-vivo rhinovirus infection. Their dysregulated genes and pathways were then compared to those of 9 human subjects prior and after intranasal inoculation in-vivo with rhinovirus. Additionally, we developed the Similarity Venn Diagram, a novel visualization method that goes beyond conventional overlap to show the similarity between two sets of qualitative measures. RESULTS: We evaluated the individual N-of-1-pathways results using two established cohort-based methods: GSEA and enrichment of differentially expressed genes. Similarity Venn Diagrams and individual patient ROC curves illustrate and quantify that the in-vivo dysregulation is recapitulated ex-vivo both at the gene and pathway level (p-values<0.004). CONCLUSION: We established the first evidence that an interpretable dynamic transcriptome metric, conducted as ex-vivo assays for a single subject, has the potential to predict individualized response to infectious disease without the clinical risks otherwise associated to in-vivo challenges. These results serve as a foundational work for personalized "virograms".

Project description:Understanding individual patient host-response to viruses is key to designing optimal personalized therapy. Unsurprisingly, in vivo human experimentation to understand individualized dynamic response of the transcriptome to viruses are rarely studied because of the obvious limitations stemming from ethical considerations of the clinical risk.In this rhinovirus study, we first hypothesized that ex vivo human cells response to virus can serve as a proxy for otherwise controversial in vivo human experimentation. We further hypothesized that the N-of-1-pathways framework, previously validated in cancer, can be effective in understanding the more subtle individual transcriptomic response to viral infection.N-of-1-pathways computes a significance score for a given list of gene sets at the patient level, using merely the 'omics profiles of two paired samples as input. We extracted the peripheral blood mononuclear cells (PBMC) of four human subjects, aliquoted in two paired samples, one subjected to ex vivo rhinovirus infection. Their dysregulated genes and pathways were then compared to those of 9 human subjects prior and after intranasal inoculation in vivo with rhinovirus. Additionally, we developed the Similarity Venn Diagram, a novel visualization method that goes beyond conventional overlap to show the similarity between two sets of qualitative measures.We evaluated the individual N-of-1-pathways results using two established cohort-based methods: GSEA and enrichment of differentially expressed genes. Similarity Venn Diagrams and individual patient ROC curves illustrate and quantify that the in vivo dysregulation is recapitulated ex vivo both at the gene and pathway level (p-values?0.004).We established the first evidence that an interpretable dynamic transcriptome metric, conducted as an ex vivo assays for a single subject, has the potential to predict individualized response to infectious disease without the clinical risks otherwise associated to in vivo challenges. These results serve as a foundational work for personalized "virograms".

Dataset Information

Deep Morphology Learning Enhances Ex Vivo Drug Profiling-Based Precision Medicine.

Significance

Publications

Deep Morphology Learning Enhances Ex Vivo Drug Profiling-Based Precision Medicine.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets