Optical tissue clearing and machine learning can precisely characterize extravasation and blood vessel architecture

Serhii Kostrikov1, Kasper B Johnsen1, Thomas H Braunstein2

  • 1Section for Biotherapeutic Engineering and Drug Targeting, Department of Health Technology, Technical University of Denmark, Lyngby, Denmark.

Insights

Transcardial perfusion overestimates drug extravasation in brain tumors due to incomplete removal from tumor vasculature. New 3D deep imaging and machine learning methods precisely quantify drug extravasation and tumor angioarchitecture.

Area of Science:

  • Neuroscience
  • Pharmacology
  • Biomedical Imaging

Background:

  • Accurate quantification of drug accumulation in brain tissue is crucial for developing therapeutics for brain disorders.
  • Transcardial perfusion is a standard method to assess drug extravasation into the brain by removing intravascular drug fractions.
  • The efficiency of transcardial perfusion can be compromised by pathological changes in the tumor microenvironment, a factor often overlooked.

Purpose of the Study:

  • To investigate the impact of tumor vasculature on transcardial perfusion efficiency.
  • To develop advanced imaging and analysis techniques for precise quantification of drug extravasation in brain tumors.
  • To characterize brain tumor angioarchitecture and its correlation with drug extravasation patterns.

Main Methods:

  • Utilized 3D deep imaging of optically cleared tumor samples to overcome limitations of traditional perfusion methods.
  • Developed two machine learning-based semi-automated image analysis workflows.
  • Quantified compound extravasation patterns and tumor angioarchitecture in large 3D datasets.

Main Results:

  • Demonstrated that transcardial perfusion is insufficient for removing injected compounds from tumor vasculature, leading to significant overestimation of extravasation compared to healthy vasculature.
  • Successfully characterized compound extravasation patterns and tumor angioarchitecture using the developed 3D imaging and machine learning workflows.
  • Established a methodology for precise and comprehensive analysis of extravasation in brain tumors.

Conclusions:

  • Transcardial perfusion is unreliable for quantifying drug extravasation in brain tumors due to incomplete intravascular washout.
  • 3D deep imaging combined with machine learning offers a precise method for analyzing drug extravasation and tumor vasculature.
  • This approach enables detailed correlation between extravasation patterns and the heterogeneous features of brain tumor vasculature, advancing therapeutic development.

Related Concept Videos