Histological assessment of invasive breast cancer stroma maturity: characterisation, prognostic significance and relationship with non-invasive radiological imaging signal

Breast cancer is the second most common cancer worldwide and the most frequent cancer among women. Breast cancer screening programs have increased early detection in asymptomatic women, however the early detection of breast lesions has led to an increase in overdiagnosis, overtreatment and unsuccessful treatment response. There is a need for improved tumour characterisation and phenotyping so that accurate prognostics and treatment plans can be tailored on a patient basis. The tumour microenvironment plays a crucial role in tumour progression and patient survival. During tumour progressing, the remodelling of collagen fibres gives rise to observable patterns in the tumour associated stroma in histological slides from clinical cases of invasive breast carcinoma that the pathologist can categorise as mature or immature. The work presented in this thesis focuses on the histological analysis of breast cancer stroma and its potential value for patient prognosis and interpretation of non-invasive radiological imaging. It is presented a framework for the automated classification of stromal regions by using a combination of Derivative-of-Gaussian filters and local binary patterns to determine the maturity of breast stroma. The results show that this classification agrees with that of skilled observers, hence providing a repeatable, quantitative and reproducible measure for prognostic studies. The framework was applied to whole-slide histological sections, which provided a way to investigate stroma heterogeneity, tumourstroma ratio and correlation with available patient clinical information, such as tumour grade and patient survival. Diffusion MRI provides information on tissue environment that can be measured in-vivo and the framework was also compared with this signal. Stroma maturity was found to be significantly correlated with tumour grade (p = 0.016) and high stromal content was associated with better survival, although the association was not significant (p = 0.10). A pilot study on the relationship between stroma maturity and the apparent diffusion coefficient from magnetic resonance imaging (MRI) of fixed samples was undertaken and results showed, for the limited data available, that mature regions revealed higher apparent diffusion coefficient (ADC) values compared to immature regions