Personalised Medicine

Future architecture

Timothy Yeatman

Executive Vice President, Translational Research, H. Lee Moffitt Cancer Center & Research Institute, University of South Florida and President & CSO M2Gen, USA


The standard “one size fits all” approach of treating many individuals may soon become obsolete. More targeted approaches promise to improve outcomes while reducing toxicity and medical costs.


Cancer is a molecularly heterogeneous disease. Simply put, not all cancers, even when derived from an organ site such as colon or lung, are alike. Despite this recently determined finding, treatments are assigned to tumour types primarily based on their site of origin. Thus, while there may be many different molecular subtypes of lung cancer, all adenocarcinomas of the lung are treated with the same chemotherapeutic agents.

Cancer therapy as we know is effective for some patients, but for others it can be toxic and has no survival benefits. Despite many therapeutic choices, still very few patients with metastatic disease are cured. Couple this fact with the clear reduction in the availability of new drugs for testing and it becomes obvious that a radical change in the drug development process is necessary in order to improve outcomes. We believe the standard “one size fits all” approach of treating many individuals may soon become obsolete. More targeted approaches promise to improve outcomes while reducing toxicity and medical costs.

We and others have answered the challenge that human tumours might be classified using a new molecular tool—the microarray. With this tool, we are now able to assess the expression of ~30,000 genes in a single day across numerous tumours, a quantum leap in the technology of mRNA-based gene expression profiling. This technological advance has made it possible to develop large data sets containing both gene expression data as well as clinical outcome and response data.

Initial studies clearly demonstrated the potential to predict diagnosis and prognosis for a number of tumour types more comprehensively than had been possible with previously available semi-quantitative immunohistochemical tools. It has become clear that no two tumours are precisely identical, with a significant amount of biological heterogeneity between and within tumours. Correlative studies at multiple sites have found that the biological variability from one tumour to the next exceeds the inherent variability or reproducibility of the test. Moreover, the potential to predict response or non-response to chemotherapy has been recently demonstrated by a number of investigators, suggesting that there could be a clinical application for this technology. Collectively, the data suggested that there might be a long-term benefit in evaluating every tumour possible using microarray technology (“one tumour, one chip”) to fully characterise the tumours' individual signatures.

It was then not a stretch to start to envision a data repository for tumour and clinical data that might be useful for a host of opportunities from target and pathway identification, to signature generation. Thinking that a database might be more valuable with data from both primary tumour and metastases, we began to devise a mechanism by which patients with metastatic diseases might draw value from this project.

The concept of “population-based trial matching” was developed, which is a clear departure from our standard means of identifying patients for therapeutic clinical trials. With a large database composed of molecular fingerprints from thousands of patients (with metastatic disease), it was clear that it might be feasible to match the right patient to the right drug in an expedient fashion. And not only would the trial be completed in record time, we hypothesise the response rates will climb due to the selective process for the identification of patient candidates. So, the future of clinical trials will be much like organ transplantation where large sophisticated computerised data systems and networks are used to find the right organ donor for the right transplant recipient.

The H. Lee Moffitt Cancer Center has begun to collaborate with multiple partners, both in academia and in industry, to develop a clinical and gene expression database for scientific research and for translational research. This database is part of a larger initiative at the Moffitt Cancer Center called “Total Cancer Care”. Total Cancer Care is a Center-wide and State-wide initiative to improve the quality of medicine and the standard of care by developing personalised approaches to cancer care whereby the best therapeutics are delivered to the patients who might benefit the most. In Total Cancer Care, we will scrutinise outcomes and survivorship. We will try to determine what barriers are there to clinical trial accrual. We will also try to actually deliver personalised cancer care back to the patient through population-based trial matching. This is truly an enterprise project that spans the State of Florida and beyond, attempting to bring new value to the participating patients, physicians and hospitals.

We have overlaid an all digital IT approach to collecting and sorting the clinical data that will be collected for the life of the patient on top of a sophisticated data warehouse that can collect, sort and relate data from many different electronic feeds and types. The data warehouse will contain data from gene expression experiments and will link this data by unique identifiers to clinical outcomes data such as overall survival and disease recurrence.

While initial pilot projects have been successful, the real challenges lie ahead when we begin to build the front end to the warehouse that will enable patients, physicians and basic researchers to access the data and process it. Thus, the future of personalised cancer care depends on our ability to operationalise a network of hospitals, physicians and nurses to collect the tissues and associated clinical data longitudinally over time. More importantly, our capacity to reach out to the patients with metastatic disease and align them with the best trial opportunity through gene profiling is critical to the success of this project.

There is more to the personalised medicine project than gene expression data sets. For example, because we plan to acquire thousands of tumour and blood samples, we will be able to interrogate these samples with other novel technologies as they are developed. We fully anticipate the potential to evaluate thousands of tumour samples for somatic gene mutations in the very near future. This will permit the development of new dimensions to the data warehouse, ultimately allowing scientists to better understand the relationships between gene expression and the underlying genetic codes and associated mutational flaws.

Beyond the development of a data warehouse and trial matching capabilities, we believe there is a great need to develop molecular imaging technologies.

The capacity to image the metabolic activities in a tumour is now becoming a reality. While currently we can examine the glucose metabolism of a tumour using 64 slice PET-CT scanners, we plan to develop imaging tools based on tumour biological endpoints such as apoptosis, proliferation, and angiogenesis. This sort of technology would permit, for the first time, the potential to measure the response of a tumour to a drug or to radiotherapy within minutes of delivery. This would be a radical change from current practice where drug responses are evaluated only after ~3 months of therapy using Response Evaluation Criteria in Solid Tumours (RECIST) that physically measure tumour diameters with CT scans. Such an approach, when integrated into a personalised medicine paradigm, might allow a rapid, iterative, reevaluation/reassignment of therapy following initial therapeutic selection and delivery.

The path to personalised medicine is neither short nor straight. We believe, however, that we have outlined a rational roadmap to deliver personalised cancer care to patients within 5-10 years. This roadmap requires precise execution of a large translational research project we call “Total Cancer Care”, that will build a research data warehouse relating clinical and molecular data in a format useful to patients, physicians and scientists.