Advanced diffusion-weighted imaging of the pancreas : pre-clinical and clinical developments
Year: 2025
Type: dissertation
Abstract: Pancreatic cancer is one of the most lethal cancers worldwide, with a higher burden in the western world. The overall survival rate at 5 years is estimated at 13%. This dismal scenario is attributed to its common late diagnosis at advanced stages, precluding surgical resection, which remains the only potentially curative option. Thus, early diagnosis is key for improving patient prognosis. Recent data established that pancreatic cancer carcinogenesis follows a progressive development of pre-malignant lesions, the most relevant being pancreatic intraepithelial neoplasia (PanIN). This process entails potential opportunities for early diagnosis and intervention, which could effectively reduce the burden of pancreatic cancer. Current imaging methods, namely computerized tomography (CT) and magnetic resonance imaging (MRI), are well established for the diagnosis and characterization of pancreatic cancer. These methods are, nevertheless, ineffective for diagnosing pre-malignant lesions, i.e., PanIN. Current imaging is also lacking when assessing response to neoadjuvant therapy for pancreatic cancer, a crucial point for patient management. Diffusion-weighted imaging (DWI) has become a staple of MRI for a great number of clinical applications, due to its microstructure probing ability, with its robust metric -apparent diffusion coefficient (ADC) being widely accepted. However, the full complexity of diffusion is not captured by conventional DWI. More advanced diffusion-based methods include diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI). DTI provides a rotationally invariant representation of the diffusion process, allowing characterization of anisotropy and diffusivities across major axes. This technique is routinely used for neuroimaging, namely for tractography, but not for pancreatic imaging. DKI explores the non-Gaussianity of diffusion, which has been associated with microstructural complexity, therefore presenting potential for characterizing pancreatic changes at the microscopic level. The works presented in this Thesis were aimed at developing tools with potential for improving the dismal reality of pancreatic cancer, through the detection of the major pre-malignant lesions – PanIN, and better characterization of response to neoadjuvant therapy. First, since DWI’s role on evaluating response to neoadjuvant therapy is not yet clear, a systematic review and meta-analysis was performed for he published literature on this matter. Then, a pre-clinical study was conducted, using genetically engineered mouse models (GEMM) which develop PanIN and pancreatic cancer. In the first part of this pre-clinical study, we developed a new method for bowel motion reduction using hyoscine butylbromide with different dosages, allowing for high-quality MRI of the mouse abdomen. Secondly, the GEMM were imaged periodically with anatomical T2-weighted pulse sequence, to determine age of onset of detectable pancreatic changes. The third part of this study consisted of developing a DTI application for in vivo GEMM, searching for PanIN and PDAC. Following in vivo DTI, we performed ex vivo MR Microscopy through ultrahigh resolution DTI, which allowed a direct correlation with histology and determination of underlying histological features responsible for MRI contrasts. As a final part of the pre-clinical study, human pancreatic samples were imaged, to evaluate the technique’s translational potential. The final study included in this Thesis was aimed at providing an adaptation of DTI and DKI of the healthy pancreas in a clinical setting, exploring its feasibility, repeatability, quality and inter-reader agreement. In the matter of DWI’s role for evaluating response to neoadjuvant therapy, our systematic review and meta-analysis revealed that, although DWI is reported as useful for this purpose, further studies with robust data are needed to provide specific recommendations for clinical practice. This conclusion was motivated by high heterogeneity and risk for bias between studies. Also, ADC values overlapped for responders and non-responders between studies, underlining the need for standardization in image acquisition and evaluation. Our first pre-clinical study demonstrated the efficacy of hyoscine butylbromide at high dosage for reducing bowel motion artifacts in abdominal MRI of the mouse. The bowel motion reduction window was observed 8.5 min after injection and lasted up to 45 min. Its use allowed the acquisition of high-quality images of the mouse abdomen, including DWI. We found that DTI can be used to detect and characterize PanIN in vivo in GEMM, with high diagnostic performance (up to AUC = 1). Using MR Microscopy, we observed specific microstructural features of PanIN and PDAC with direct histological correlation: anisotropy in peripheral areas of fibrosis, high diffusivities in cystic areas and high axial diffusivity and T2* in stromal components. These same contrasts were present in the human pancreas allowing the identification of PanIN, determining its potential for clinical translation. Finally, we developed a clinically feasible DTI and DKI acquisition for pancreatic imaging. Both techniques were acquired in a single session, with excellent inter-observer agreement. Although DTI provided optimal image quality and repeatability, MK inaccuracies were observed, prompting further research into artifacts correction. A shorter protocol with 6 diffusion directions may be preferred. In conclusion, the works presented in this Thesis have contributed to furthering knowledge on pancreatic cancer imaging. The most relevant findings for clinical practice include the clarification of DWI’s role in evaluating response to neoadjuvant therapy, and the development of clinically feasible DTI and DKI acquisition of the pancreas, with high-quality DTI. The development of DTI for the detection and characterization of PanIN offers new possibilities for researching pancreatic cancer carcinogenesis, both in pre-clinical and clinical contexts. Simultaneously, our findings represent a first step towards early and noninvasive detection of PanIN, which could ultimately provide more effective avenues for treating, or even preventing, pancreatic cancer.
Author: Bilreiro, Carlos
Language: Portuguese
Topic: MRI in cancer diagnosis
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Keywords: diffusion-weighted imaging, pancreatic cancer, pancreatic intraepithelial neoplasia, diffusion kurtosis imaging, diffusion tensor imaging, apparent diffusion coefficient, neoadjuvant therapy response, hyoscine butylbromide, genetically engineered mouse models, pancreatic ductal adenocarcinoma, early diagnosis
Open Access status: green
Funder: Fundação Champalimaud
OpenAlex ID: W7079911722
