Abstract
BackgroundMagnetic resonance imaging (MRI) has gained a key role in the management of cancer patients due to high-resolution depiction of soft tissue anatomy. Novel MRI biomarkers that inform on tumour biology, physiology, and their changes in response to therapy can be obtained from state-of-the-art clinical scanners and require clinical validation.
Overall Hypothesis
The physiology and biomechanics of primary abdominal cancers, including renal cell carcinoma, colorectal adenocarcinoma, and squamous anal carcinoma, can be measured by means of clinical MR imaging. Novel biomarkers derived from MRI Elastography, oxygen-enhanced MRI and diffusion-weighted imaging can undergo initial validation in the clinical setting.
Objectives
1. Assess the feasibility of 3T MRI elastography (MRE) as a biomarker ofmalignancy in healthy volunteers and in patients with small renal masses(Sections 2.2 and 2.3). We hypothesized that the viscoelastic shear propertiesof renal cell carcinoma measured by MRE would reflect the underlyinghistopathological structure and differ from benign oncocytoma, oftenindistinguishable based on conventional anatomical imaging.
2. Assess the feasibility of oxygen-enhanced MRI (OE-MRI) as a biomarker ofhypoxia in patients with primary colorectal cancer (Section 3.3). Tumourhypoxia is an important adverse prognostic factor. Changes in the effectivetransverse relaxation time (ΔT2*) and longitudinal relaxation time (ΔT1)induced by respiratory oxygen challenge have shown promise as biomarkers of tumour oxygenation in preclinical colorectal cancer models and are yet to bevalidated clinically.
3. Investigate the impact of diffusion-weighted imaging (DWI) on tumour staging, tumour volume delineation and response assessment in patients withsquamous cell carcinoma of the anal canal (SCCA) (Sections 4.2 and 4.3). DWI isestablished clinically as a valuable complement to T2-weighted sequences inprimary staging and response assessment of rectal adenocarcinoma but has notbeen evaluated in SCCA.
Methods
This thesis includes two prospective and two retrospective clinical studies:
1. Ten healthy volunteers and 21 patients with small renal masses were enrolledprospectively. Three-directional, gradient echo 3T MRE was employed.Viscoelastic parametric maps (shear wave velocity and attenuation) wereassessed by two independent radiologists. Within-subject repeatability andinterobserver agreement were evaluated. Imaging parameters were comparedbetween patients with renal cell carcinoma (RCC) and benign renal oncocytoma(Sections 2.2 and 2.3).
2. Twenty-two patients with primary colorectal adenocarcinoma were enrolledprospectively and imaged at 3T. T1 and T2* maps were obtained using singlesliceand 3D multi-slice techniques, applied during air breathing and duringpure oxygen breathing. Tumour ΔT1 and ΔT2* were measured by twoindependent radiologists. Within-subject repeatability of T1 and T2*measurements and interobserver agreement were assessed (Section 3.3).
3. The staging MRI scans of 45 patients with SCCA were assessed retrospectivelyby two independent radiologists. Mean tumour diameter (MTD) and grosstumour volume (GTV) were delineated on T2-weighted and high-b-value DWIimages, and compared between sequences. Tumour staging was derived fromMTD. Interobserver agreement was assessed, and delineation confidencescored (1 to 5) by each observer (Section 4.2).
4. The staging and post-treatment MRI scans of 85 patients with SCCA wereassessed retrospectively by two experienced and two less experiencedindependent radiologists. Tumour regression grading (TRG) score (1-5) was firstassigned using triplanar T2-weighted sequences alone; a month later, usingcombined T2-weighted sequences and DWI (b≥800). The number ofindeterminate (TRG-3) scores obtained from the two separate reading methodswas compared. Observer confidence was graded (1-5) after each TRG score.Interobserver agreement was assessed (Section 4.3).
Results
1. Within-subject repeatability for MRE was acceptable, with coefficients ofvariation of 8-12%. Oncocytomas displayed a consistent MRE viscoelasticprofile, characterised by significantly lower shear velocity and higher shearattenuation than clear cell RCC. MRE interobserver agreement was excellent.
2. Single-slice modified Look-Locker inversion recovery sequences (MOLLI) yielded tumour T1 measurements of sufficient quality in 12 participants. All other sequences were degraded by motion and susceptibility artefact. A statistically significant negative ΔT1 of approximately 2% was observed in 8/12 tumours, in line with the expected response in well perfused and oxygenated tissue. Within-subject repeatability for MOLLI T1 measurements was affected bymotion, with coefficients of variation of 11-17%. Interobserver variability was<10%.
3. GTV and MTD were significantly and systematically lower on DWI than on T2-weighted sequences, causing tumour (T) staging discordances in a quarter ofcases. Interobserver agreement was higher for DWI. Delineation confidencewas greater on DWI.
4. The number of indeterminate TRG-3 scores decreased significantly bycomplementing T2-weighted sequences with DWI. For all observers combined,the number of TRG-3 cases halved from 24% of the total to 12%. Most of theindeterminate TRG-3 scores (82%) resolved in favour of excellent response(TRG-2). Interobserver agreement was fair to moderate and did not changesignificantly between imaging methods. Confidence in assessing responseincreased significantly with the addition of DWI for all observers.
Conclusion
MRI biomarkers of cancer physiology (water diffusion) and cancer biomechanics(viscoelasticity) can be measured effectively by means of clinical MR imaging.Knowledge acquired from three clinical studies shows readiness for qualification in prospective clinical trials, and potential for patient benefit. MRI biomarkers of tumour hypoxia in colorectal cancer require further technical development and evaluation before they can be implemented in clinical trials.
| Date of Award | 1 May 2023 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Vicky Goh (Supervisor) & Gary Cook (Supervisor) |