Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth
<strong>Introduction:</strong> As there is no viable medical treatment to halt the progression of Abdominal Aortic Aneurysms (AAA), current research focus is on the identification of novel methods to predict AAA growth, allowing for risk stratification and timing of surgical intervention...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: |
| _version_ | 1797113159227539456 |
|---|---|
| author | Ngetich, EK |
| author2 | Handa, A |
| author_facet | Handa, A Ngetich, EK |
| author_sort | Ngetich, EK |
| collection | OXFORD |
| description | <strong>Introduction:</strong> As there is no viable medical treatment to halt the progression of Abdominal Aortic Aneurysms (AAA), current research focus is on the identification of novel methods to predict AAA growth, allowing for risk stratification and timing of surgical intervention. This DPhil thesis aimed to i) assess the performance of radiomic features in predicting AAA growth and systemic endothelial function, ii) investigate the metabolomic fingerprint of patients with AAA and the performance of a metabolic signature in predicting future AAA growth, iii) investigate metabolic pathways and novel biomarkers for AAA progression. <strong>Methods:</strong> Radiomics analysis utilized CT scans from a retrospective multicohort sample. Image segmentation and volume reconstruction was done using semi-automated method developed by our research group and radiomics features were extracted using Python software. Model performance was assessed using Receiver Operating Characteristic (ROC) curves and Root Mean Square Error (RMSE). Untargeted metabolomics analysis was conducted using four pairwise experiments: (i) comparison of patients with AAA vs healthy volunteers (HV), (ii)fast vs slow growing AAA, (iii) before vs after AAA open surgical repair, (iv) ILT tissue vs omental tissue. Global metabolomics analysis was done using Atmospheric Pressure Solid Analysis Probe Mass spectrometry. Prediction of future AAA growth was assessed using ROC curves. Targeted metabolomics was conducted using Ion Chromatography and C18 High Performance Liquid Chromatography. Functional metabolite relevance was assessed using the Mummichog software in Metaboanalyst 5.0 platform. <strong>Results:</strong> Radiomic signature performed marginally better at predicting fast and slow AAA growth compared to AAA diameter [slow growth (Accuracy 77.6%, AUC 0.61 vs accuracy 62.2%, AUC 0.46)] and [fast growth (Accuracy 78.8%, AUC 0.72 vs accuracy 67.3%, AUC 0.51)]. AAA diameter, however, seemed to perform better than radiomic signature at predicting endothelial function using both linear (RMSE 2.145 vs RMSE 2.723) and logistic models (Accuracy 49.3 vs 52.3). On untargeted metabolomics, univariate analysis of metabolites in the four experiments showed Differentially Expressed Metabolites (DEMs) in all four pairwise comparisons and integration using Venn diagrams revealed three metabolites m/z 177.10, m/z 191.15, m/z 362.15 that were (i) significantly elevated in AAA compared to HV, (ii) elevated in fast compared to slow AAA, (iii) increased in AAA patients before surgery compared to after surgery and finally, (iv) elevated in ILT tissue compared to the omental tissue. The metabolic signature performed better than AAA diameter at predicting future growth for fast-growing aneurysms AUC 77.8%, p<0.00001 vs AAA diameter of AUC 62.6%, p=0.057. Pathway-level integration analysis revealed enriched pathways including 3 pathways not previously described in association with AAA. These included glycosphingolipid biosynthesis-ganglio series, Pentose phosphate and mucin type O-glycan biosynthesis. Compound identification revealed metabolites as N-formyl- methionine and citric acid as possible biomarkers of AAA growth. <strong>Conclusion:</strong> Radiomic and metabolic features correlate with AAA growth and has potential to predict AAA growth. Metabolomic biomarkers offer promise in the search for robust AAA biomarkers and open a wide field for further research and validation of the identified compounds and metabolic pathways.
|
| first_indexed | 2024-04-23T08:24:23Z |
| format | Thesis |
| id | oxford-uuid:99a1b897-a051-49d3-92e2-9a5f1580c6a1 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-04-23T08:24:23Z |
| publishDate | 2023 |
| record_format | dspace |
| spelling | oxford-uuid:99a1b897-a051-49d3-92e2-9a5f1580c6a12024-04-10T14:55:47ZInvestigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growthThesishttp://purl.org/coar/resource_type/c_db06uuid:99a1b897-a051-49d3-92e2-9a5f1580c6a1Blood-vessels--DiseasesAbdominal aneurysmCardiovascular systemEnglishHyrax Deposit2023Ngetich, EKHanda, ARegent, LKokila, LFrank, SClaire, V<strong>Introduction:</strong> As there is no viable medical treatment to halt the progression of Abdominal Aortic Aneurysms (AAA), current research focus is on the identification of novel methods to predict AAA growth, allowing for risk stratification and timing of surgical intervention. This DPhil thesis aimed to i) assess the performance of radiomic features in predicting AAA growth and systemic endothelial function, ii) investigate the metabolomic fingerprint of patients with AAA and the performance of a metabolic signature in predicting future AAA growth, iii) investigate metabolic pathways and novel biomarkers for AAA progression. <strong>Methods:</strong> Radiomics analysis utilized CT scans from a retrospective multicohort sample. Image segmentation and volume reconstruction was done using semi-automated method developed by our research group and radiomics features were extracted using Python software. Model performance was assessed using Receiver Operating Characteristic (ROC) curves and Root Mean Square Error (RMSE). Untargeted metabolomics analysis was conducted using four pairwise experiments: (i) comparison of patients with AAA vs healthy volunteers (HV), (ii)fast vs slow growing AAA, (iii) before vs after AAA open surgical repair, (iv) ILT tissue vs omental tissue. Global metabolomics analysis was done using Atmospheric Pressure Solid Analysis Probe Mass spectrometry. Prediction of future AAA growth was assessed using ROC curves. Targeted metabolomics was conducted using Ion Chromatography and C18 High Performance Liquid Chromatography. Functional metabolite relevance was assessed using the Mummichog software in Metaboanalyst 5.0 platform. <strong>Results:</strong> Radiomic signature performed marginally better at predicting fast and slow AAA growth compared to AAA diameter [slow growth (Accuracy 77.6%, AUC 0.61 vs accuracy 62.2%, AUC 0.46)] and [fast growth (Accuracy 78.8%, AUC 0.72 vs accuracy 67.3%, AUC 0.51)]. AAA diameter, however, seemed to perform better than radiomic signature at predicting endothelial function using both linear (RMSE 2.145 vs RMSE 2.723) and logistic models (Accuracy 49.3 vs 52.3). On untargeted metabolomics, univariate analysis of metabolites in the four experiments showed Differentially Expressed Metabolites (DEMs) in all four pairwise comparisons and integration using Venn diagrams revealed three metabolites m/z 177.10, m/z 191.15, m/z 362.15 that were (i) significantly elevated in AAA compared to HV, (ii) elevated in fast compared to slow AAA, (iii) increased in AAA patients before surgery compared to after surgery and finally, (iv) elevated in ILT tissue compared to the omental tissue. The metabolic signature performed better than AAA diameter at predicting future growth for fast-growing aneurysms AUC 77.8%, p<0.00001 vs AAA diameter of AUC 62.6%, p=0.057. Pathway-level integration analysis revealed enriched pathways including 3 pathways not previously described in association with AAA. These included glycosphingolipid biosynthesis-ganglio series, Pentose phosphate and mucin type O-glycan biosynthesis. Compound identification revealed metabolites as N-formyl- methionine and citric acid as possible biomarkers of AAA growth. <strong>Conclusion:</strong> Radiomic and metabolic features correlate with AAA growth and has potential to predict AAA growth. Metabolomic biomarkers offer promise in the search for robust AAA biomarkers and open a wide field for further research and validation of the identified compounds and metabolic pathways. |
| spellingShingle | Blood-vessels--Diseases Abdominal aneurysm Cardiovascular system Ngetich, EK Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title | Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title_full | Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title_fullStr | Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title_full_unstemmed | Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title_short | Investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm (AAA) growth |
| title_sort | investigating radiomic and metabolomic biomarkers of abdominal aortic aneurysm aaa growth |
| topic | Blood-vessels--Diseases Abdominal aneurysm Cardiovascular system |
| work_keys_str_mv | AT ngetichek investigatingradiomicandmetabolomicbiomarkersofabdominalaorticaneurysmaaagrowth |