Fast and sensitive dynamic oxygen‐enhanced MRI with a cycling gas challenge and independent component analysis

• Published in: Magnetic resonance in medicine Vol. 81; no. 4; pp. 2514 - 2525
• Main Authors: Moosvi, Firas, Baker, Jennifer H.E., Yung, Andrew, Kozlowski, Piotr, Minchinton, Andrew I., Reinsberg, Stefan A.
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.04.2019
• Subjects: Animal models; Animals; Biomarkers; Biomarkers - metabolism; Biomarkers, Tumor - metabolism; Blood vessels; Cell Hypoxia; Cell Line, Tumor; Change detection; Cycles; Data processing; Female; HCT116 Cells; Histology; Humans; Hypoxia; Image acquisition; Image detection; Image Processing, Computer-Assisted - methods; Independent component analysis; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Mice; Neoplasm Transplantation; Neoplasms - diagnostic imaging; Nitroimidazoles - therapeutic use; Oxygen; Oxygen - metabolism; oxygen enhanced MRI; Oxygenation; Platelet Endothelial Cell Adhesion Molecule-1 - metabolism; Respiration; Response functions; Sensitivity; Sensitivity analysis; Tumor Microenvironment; Tumors; hypoxia; tumors; oxygenation; oxygen enhanced MRI
• ISSN: 0740-3194; 1522-2594
• DOI: 10.1002/mrm.27584

Purpose There is a critical need for non‐invasive imaging biomarkers of tumor oxygenation to assist in patient stratification and development of hypoxia targeting therapies. Using a cycling gas challenge and independent component analysis (ICA), we sought to improve the sensitivity and speed of existing oxygen enhanced MRI (OE‐MRI) techniques to detect changes in oxygenation with dynamically acquired T1W signal intensity images (dOE‐MRI). Methods Mice were implanted with SCCVII, HCT‐116, BT‐474, or SKOV3 tumors in the dorsal subcutaneous region and imaged at 7T. T1W images were acquired during a respiratory challenge with alternating 2‐minute periods of air and 100% oxygen for three cycles. Data were analyzed with ICA and oxygenation maps were generated and compared to corresponding histology sections stained for hypoxia (pimonidazole) and blood vessels (CD31). Results Cycling air‐oxygen‐air gas challenges were well tolerated and ICA permitted extraction of the oxygen‐enhancing component in all imaged tumors from four different models. Comparison with synthetic response functions showed that dOE‐MRI does not require any a‐priori knowledge of the physiological response. The fraction of O2‐negative dOE‐MRI voxels that correlate inversely with the ICA gas‐cycling component correspond well with the histological hypoxic fraction in SCCVII tumors (r = 0.91, p = 0.0016) but did not correlate in HCT‐116 tumors (r = 0.13, p = 0.81). Conclusions Using ICA and adding a cycling gas challenge extends the sensitivity of OE‐MRI and allows the oxygenation status of tumors to be assessed in as little as six minutes. These findings support further development of OE‐MRI as a biomarker of tumor oxygenation.