This research, led by Professor Graham Ball, has developed new bioinformatics techniques for mining complex post genomic bio-profile data. The approach allows the development of predictive models to answer clinical questions using an optimum biomarker panel. The findings of the research mean that the prognosis for women with breast cancer could be better predicted in future. The impact of this work is demonstrated in the filing of four patents associated with algorithms, breast cancer and tuberculosis, subsequently licensed to a spin-out company. Three clinical trials have been supported and others are in the pipeline. Through the spin-out company the technique is being applied to stratify patients in clinical collaborations and to optimise biomarker panels for diagnostics companies.
The area of biomarker discovery has seen significant developments over the last 10 years and the team at NTU continues to offer unique and leading approaches in the field. Novel non-linear approaches have been applied to the identification of biomarkers, from complex genomic data, addressing clinical questions such as prognosis and response to therapy. These biomarkers have subsequently been validated using immunohistochemical techniques and applied in clinical practice and decision making.
Clinical collaboration with Nottingham University Hospitals Trust has identified biomarkers of proliferation and prognosis in breast cancer. These have been used to re-define the Nottingham Prognostic Index (NPI) allowing a more accurate prediction of prognosis for the individual. The most influential marker predicts response to Anthracycline and the set has been successfully evaluated as predictive using immunohistochemistry. This approach has been used by multi-disciplinary teams in clinical decision making and to evaluate prognosis in medico-legal cases, and it has influenced decisions around patient care and stratification. In addition, the approaches developed have been used in identifying markers associated with a wide range of clinical groups across further clinical conditions.
This biomarker discovery work has led to the filing of four patents:
- Data Analysis Method and System
- Time to Event Data Analysis Method & System
- TB Marker
- SPAG 5 Biomarker
These patents are licenced to CompanDX Ltd. Further patents are currently under development for pancreatic cancer, cardiovascular disease and Alzheimer’s disease. Professors Ball and Robert Rees are founders and directors of this company.
CompanDX has secured significant contracts from large pharmaceutical and diagnostics companies to use the bioinformatics technologies for personalised medicine in international clinical trials. These biomarker discovery approaches have impacted on trial design and have increased the efficiency of diagnostic development, thus making trials and diagnostics more cost effective.
The patents have resulted in investment into CompanDX in partnership with New Summit Biopharma Co Ltd. As a result of this investment, clinical trials are underway in China to evaluate the efficacy of diagnostics for time-to-event in breast cancer, and diagnosis of tuberculosis. This clinical trial will validate the biomarkers in the context of regulatory approval.
The underpinning research was led by Professor Graham Ball (Professor of Bioinformatics) in collaboration with Professor Robert Rees (Director of the John van Geest Cancer Research Centre) over 12 years, focusing on bioinformatics algorithms for biomarker discovery. Professor Ball has been involved in a number of clinical collaborations leading to the publication of 72 papers since 2002, 50 of which have directly utilised the algorithms developed.
The technique consists of algorithms based on artificial neural networks (ANNs) that facilitate analysis of complex biological data, such as mass spectrometry, gene expression arrays and miRNA arrays. One of the problems with analysis of such data is its complexity and dimensionality. This leads to over-fitting and false discovery. The algorithms developed by the Ball group overcome these problems through extensive cross-validation coupled with biomarker selection based on a stepwise additive approach. Numerous publications by the group, for example this 2009 paper, demonstrate that these limitations have been overcome. The algorithms have been applied to the analysis of clinical data to identify an optimised subset of markers and incorporate them into a model that best predicts an answer to a given clinical question. These markers and models provide an insight into disease aetiology and can be used from a diagnostic perspective.
The group was one of the first to mine mass spectrometry data using an ANN approach. The initial study identified biomarker ions that accurately differentiated between astrocytoma and glioblastoma. This study showed that it was possible to use the non-linear predictive capabilities of ANNs to classify a clinical state using biomarker ions from SELDI-MS data. These approaches were then developed further, and applied to analysis of melanoma data for a larger cohort. Subsequently, the methods were refined, developed and validated to improve performance, optimise the biomarker panels identified and overcome limitations associated with high dimensionality and the complexity of the data. These new methods facilitated improved predictive performance for unseen cases from disease populations.
The approach has been applied:
- in a prostate cancer vaccine clinical trial in 2005. This study demonstrated that a cytokine profile derived using an ANN model could stratify the response of patients on a clinical trial with high sensitivity and specificity (Onyvax Ltd).
- to integrate immuno-histochemical data, pathological data, gene expression and miRNA data (Habashy et al, 2008. European Journal of Cancer, 44:11, 1541-1551.) (Lowery et al, 2009. Breast Cancer Research 11:R27.)
- to the rapid typing of microbial pathogens from mass spectrometry data, in collaboration with Public Health England.
- in the characterisation of breast cancer – contributing to the revision and remodelling of the Nottingham Prognostic Index.
- Nottingham University Hospitals Trust, Breast Cancer Pathologist. Will corroborate the clinical impact of breast cancer biomarkers and the use of such biomarkers in the development of the Nottingham Prognostic Index.
- Potter Rees Ltd (serious injuries solicitors), letter of agreement available to corroborate support given in medico-legal cases where models based on the NPI have been used to predict median survival.
- Pannone Law Group, letter of agreement available to corroborate support given in medico-legal cases where models based on the NPI have been used to predict median survival.
- Public Health England, Principal Scientist. Will corroborate the impact of biomarker identification in infectious diseases, including those associated with tuberculosis.
- CompanDX, Chief Executive Officer. Will corroborate the commercial impact of the algorithms and biomarkers, particularly on projects running in China.
- Article demonstrating the funding received from China for clinical work.
- Lancashire, L. J., Powe, D. G., Reis-Filho, J. S., Rakha, E., Lemetre, C., Weigelt, B., Abdel-Fatah, T. M., Green, A. R., Mukta, R., Blamey, R., Paish, E. C., Rees, R. C., Ellis, I. O., Ball, G. R., 2010. A validated gene expression profile for detecting clinical outcome in breast cancer using artificial neural networks.Breast Cancer Research and Treatment, 120(1). 83-93.
- Ball, G., Mian, S., Holding, F., Allibone, R. O., Lowe, J., Ali, S., Li, G., McCardle, S., Ellis, I. O., Creaser, C. and Rees, R. C., 2002. An integrated approach utilizing artificial neural networks and SELDI mass spectrometry for the classification of human tumours and rapid identification of potential biomarkers. Bioinformatics, 18(3). 395-404.
- Mian, S., Ugurel, S., Parkinson, E., Schlenzka, I., Dryden, I., Lancashire, L., Ball, G., Creaser, C., Rees, R. and Schadendorf, D., 2005. Serum proteomic fingerprinting discriminates between clinical stages and predicts disease progression in melanoma patients. Journal of Clinical Oncology, 23(22). 5088-5093.
- Michael, A., Ball, G., Quatan, N., Wushishi, F., Russell, N., Whelan, J., Chakraborty, P., Leader, D., Whelan, M. and Pandha, H., 2005. Delayed disease progression after allogeneic cell vaccination in hormone-resistant prostate cancer and correlation with immunologic variables. Clinical Cancer Research, 11(12). 4469-4478.
- Lancashire, L., Schmid, O., Shah, H. and Ball, G., 2005. Classification of bacterial species from proteomic data using combinatorial approaches incorporating artificial neural networks, cluster analysis and principal components analysis. Bioinformatics, 21(10). 2191-2199.
- Blamey, R. W., Ellis, I. O., Pinder, S. E., Lee, A. H. S., Macmillan, R. D., Morgan, D. A. L., Robertson, J. F. R., Mitchel, M. J., Ball, G. R., Haybittle, J. L. and Elston, C. W., 2007. Survival of invasive breast cancer according to the Nottingham Prognostic Index in cases diagnosed in 1990-1999. European Journal of Cancer, 43(10). 1548-1555.
- The NPI plus project