UCSF

Hyperpolarized carbon-13 (HP-13C) is emerging as a powerful new molecular imaging technique utilizing specialized instrumentation and dynamic nuclear polarization (DNP) to provide a signal enhancement of over 5 orders of magnitude. HP-13C MRI has made possible quantitative detection of metabolism and perfusion in vivo with 13C-labelled biomarkers, which are safe, non-radioactive and nontoxic. Prostate cancer is the second deadliest cancer in US men, and has become a healthcare problem worldwide. A major challenge in the clinical management of prostate cancer is to determine its aggressiveness.

Through this dissertation project, a 3D dynamic compressed sensing MRSI technique was advanced from preclinical imaging to phase II clinical trial prostate cancer research. I conducted simultaneous metabolic and perfusion imaging on a transgenic mouse model of prostate cancer (TRAMP) using co-polarized 13C pyruvate and urea. Pyruvate to lactate flux (kPL) was significantly higher (p<0.001, 0.056 ± 0.005 versus 0.019 ±0.001 sec-1) for high- versus low-grade TRAMP tumors, urea AUC significantly reduced (p<0.01, 640 ± 94 versus 1407 ± 221 AU), while ktrans significantly increased (p<0.01, 358 ± 38 versus 180 ± 24 ml/dL/min). The HP-13C MRI outcomes strongly correlates with histological, gene expression and LDHA activity findings.

Translation from mice to humans requires overcoming challenges of larger imaging volume, reduce peak RF power, and decreased sensitivity. These challenges were addressed by designing new RF pulses that reduced 67% peak power, and transitioning from DSE to FID acquisitions without SNR loss. The improved sequence allows 0.5cm3 spatial and 2s temporal resolution, and enables reproducible human prostate acquisitions. “Goodness” of dynamic models was compared using Akaike’s information criteria. Quantitative accuracy was improved using a B1-insensitive variable flip angle scheme. Also investigated was the impact of pulse sequence design and parameter on kPL estimation.

New sampling patterns are proposed for larger coverage or finer resolution, and an SVD-based algorithm was applied to allow parallel reconstruction of multichannel brain data. Finally, I investigated the response of androgen deprivation therapy from a prostate cancer patient, and found substantial decrease in kPL , which indicates the clinical potential of quantitative 3D dynamic HP-13C MRI for the detection of early treatment response.