Prostate cancer (PCa) is the commonest male malignancy, with the global burden of disease expected to double by 2020. While the introduction of serum prostate-specific antigen testing has led to overdiagnosis of indolent disease, more than half of patients continue to present with locally advanced and/or metastatic PCa. Consequently, there is a pressing need to develop accurate diagnostic tools to detect and characterise clinically significant disease.

The recent international adoption of pre-biopsy magnetic resonance imaging (MRI) as the first-line investigation in patients with suspected PCa has created a paradigm shift in PCa diagnostic pathway. With its negative predictive value above 90%, pre-biopsy MRI halves the number of unnecessary biopsies and significantly reduces over-detection of indolent disease. However, MRI is poor at differentiating indolent from clinically significant lesions, thereby necessitating histopathological confirmation limited by the potential for sampling error and procedure-associated complications. Therefore, improving the diagnostic performance of standard-of-care MRI and its specificity for aggressive disease could reduce the requirement for tissue sampling and associated patient morbidity.

Imaging the metabolic alterations that occur during tumour development is a promising approach for improving the diagnostic potential of MRI. Hyperpolarised [1-13C]pyruvate MRI (HP-13C-MRI) is an emerging clinical metabolic imaging technique that can probe the exchange of the hyperpolarised 13C label between pyruvate and lactate, a biochemical reaction traditionally viewed as the metabolic hallmark of aggressive cancers. The first-in-man study of the technique showed increased [1-13C]lactate labelling in PCa areas compared to the benign prostate, with more recent studies highlighting the potential of HP-13C-MRI for tumour grade group differentiation and treatment response assessment. This dissertation builds on these early studies to advance clinical translation of HP-13C-MRI by conducting clinical and biological validation of its ability to detect and stratify PCa in patients with clinically challenging intermediate-risk disease.

To do so, Chapter 1 first introduces the key clinical concepts of PCa epidemiology, diagnosis, and risk-stratification that are directly relevant to this dissertation. It then describes the role of standard-of-care MRI in PCa diagnosis and management and highlights three key clinical decision points where HP-13C-MRI could add value to routine imaging. The chapter ends with a detailed overview of PCa metabolism and its implications for the specific methodological framework used to analyse and interpret clinical HP-13C-MRI data obtained in this dissertation.

Chapter 2 describes patient characteristics, along with imaging and biological methods used in this dissertation to assess prostate metabolism across different scales. In addition to clinical HP-13C-MRI, these methods include spatial metabolomics, immunohistochemistry, and spatial transcriptomics, with their combination critical for dissecting metabolic heterogeneity across different cellular compartments and histopathological entities.

Chapter 3 details the use of these methods for understanding the ability of HP-13C-MRI to detect PCa and differentiate biopsy-proven tumour areas from the healthy prostatic tissue. Specifically, differential [1-13C]lactate labelling between the benign and malignant prostate is interpreted in the context of both metabolic and non-metabolic properties of the two tissue types, revealing the importance of epithelial cell density and vascularity for generating HP-13C-MRI signal.

Chapter 4 uses similar methodological framework to identify clinical and biological correlates of differential [1-13C]lactate labelling between PCa of different aggressiveness, including tumours with varying percentage of Gleason pattern 4 disease and the presence of invasive cribriform component.

Chapter 5 links pre-treatment HP-13C-MRI data with intermediate-term surgical outcome data to investigate the ability of lactate imaging to detect patients at risk of developing post-surgical biochemical recurrence. Finally, Chapter 6 uses the available imaging and biological data to identify an optimal approach for HP-13C-MRI segmentation that can also be used clinically to target biopsies towards highly glycolytic tumour areas. The final section of this chapter summarises the main findings of this dissertation and provides directions for future work in the field.