Bridging the Scale Gap
: linking 2D proteomic and 3D tumour spheroid morphometric data using partial least squares structural equation modelling

  • Stefanie Wilson

    Student thesis: Doctoral Thesis

    Abstract

    Cancer is a leading health concern worldwide with approximately 17 million new cases in 2018. Treatment for cancer includes the use of chemotherapeutic drugs, radiation and surgery. Research into possible new therapeutics, and improvement of existing treatment for the different pathologies of cancer, is predominantly undertaken under laboratory conditions with the use of secondary immortalised cell lines cultured as monolayers before testing in animal models and then clinical trials in patients. There is often a disconnect between the reported efficacy of therapeutic agents in these monolayer cultures and the response in situ, often with the latter showing a reduced response compared to the former. Immortalised cell lines can be cultured as 3D aggregates of cells referred to as ‘spheroids’. These spheroid cultures often exhibit responses to therapeutic agents more similar to that of tumours and can serve to bridge the efficacy gap between 2D and tumours in situ.

    This research examines the relationships between data streams from 2D monolayer and 3D spheroid cultures through the use of statistical modelling, specifically PLS-SEM path modelling, an established technique that to the author’s knowledge, has not been applied to biological data of this kind. Data from variables concerning cell viability and proteomics in monolayer cultures and viability and morphometrics data in spheroid cultures were collected experimentally and used in a PLS-SEM path model designed for these variables. The research hypotheses explored are that: (1) variables collected from monolayer cultures correlate with variables from spheroid cultures after exposure to therapeutic insult over time: and (2) relationships between the monolayer viability and proteomic variables and spheroid morphometric indicators in responses to therapeutic insult over time could be exploited through the use of statistical modelling to predict the spheroid response from 2D data inputs.

    Results show that the novel application of PLS-SEM path modelling was successful in linking the data streams from both monolayer and spheroids. Findings of note from the evaluation of the model showed that the results of higher concentrations of 5-FU at longer time points provided stronger values of predictive accuracy and reliability in the endogenous spheroid latent variables and the pronounced mediating effect via the monolayer viability construct between the proteomics and spheroid constructs. From these findings there is scope for future work within the development of the model and with this technique more generally.
    Date of AwardOct 2019
    Original languageEnglish
    Awarding Institution
    • Abertay University
    SponsorsNorthwood Charitable Trust
    SupervisorAnne Savage (Supervisor) & Yusuf Deeni (Supervisor)

    Keywords

    • Partial least squares structural mdelling
    • Cancer
    • Spheroids

    Cite this

    '