Abstract
Dimitrios H Roukos studied Medicine at the University of Athens, Greece, and specialized in General Surgery at the JW Goethe-University Hospital in Frankfurt (1983–1989), Germany, where he received the title Dr med. Since 1990, Roukos has worked at Ioannina University School of Medicine. His current research activity includes the integration of both standard clinical and treatment data, and genetics/personal genomics/protein interactions/signaling pathways network data to predict a genotypic (QTLs) and phenotypic (cancer diversity) map. He also works to integrate systems biology, in silico, bioinformatics and reverse engineering data to achieve personalized cancer biomedicine. In the past he has published innovative models, algorithms, concepts and invited papers (including editorials and perspectives). He has also participated in keynote plenary presentations and lectures (Quest speaker international meetings). He has acheived over 1500 citations, a Hirsch factor of 28 and a total impact factor of over 600.
▪ Can you provide the readers of Expert Review of Molecular Diagnostics with an overview of your research in the area of clinical diagnostics?
Over the last few decades, advances in basic, translational and clinical research have yielded to standardization of surgery, radiotherapy and empirical chemotherapy in the multimodal treatment of solid cancers. The addition of targeted agents, despite initial enthusiasm, has modestly increased survival. As a result, cancer is, according to the current USA Cancer Facts and Statistics data, the major health problem. Estimates for the future by the WHO are disappointing. Indeed, the aging of the population, the climate changes and the worsening of the environment are all causes contributing to a rise in cancer incidence by more than 55% in the year 2025. New ideas, new conceptual scientific thoughts and changes in drug development design by the pharmaceutical industry are needed.
My research is focused on how both conventional and new evidence emerging from next-generation technology could be translated into personalized cancer management and individualized biomedicine.
▪ In your opinion, what molecular diagnostic technology has been the most successful in analyzing predispositions to cancer?
Exciting single-gene and protein-coding gene research over the last 10 years or more has revolutionized our scientific thinking and medical practice for inherited cancer. Several single genes with heritable mutations that cause cancer have been discovered. These high penetrance heritable mutations have the ability to cause cancer in approximately 75% of mutation carriers. Genetic testing is currently widely available, and can identify individuals who face a very high risk of developing breast or ovarian cancer (BRCA1 and BRCA2), colorectal or endometrial cancer (mismatch-repair genes) and diffuse gastric cancer (CDH1). For these high-risk individuals, effective interventions include starting intensive surveillance from an early age and even prophylactic surgery is now available. Current genetic testing, therefore, not only guides decisions in the primary prevention setting but it also has important implications in selecting the most appropriate extent of surgery in patients, who, in addition, are carriers of heritable mutations.
However, these mutations with large effects on cancer risk are rare: less than 0.3% of people in the general population are carriers and candidates of cancer development. Among cancer patients, they account for less than 10% of breast cancer and 3% of colorectal or gastric cancer cases that are diagnosed annually.
The plausible question, which is currently of great clinical and research importance, is what to do with the vast majority of people in the general population? How can we stratify all these individuals into high, intermediate and low risk of developing a specific cancer type? This is the decisive factor. Once this has been achieved, we can then tailor the best preventive intervention. For example, in high-risk individuals, an aggressive prophylactic surgery may be justifiable. However, for low-risk individuals, simple recommendations for physical exercise, diet and lifestyle are sufficient.
▪ To what extent do you feel that the analysis of genetic variation and gene–environment interaction can be used to assess predisposition to cancer?
It is very important to have the complete catalogue of genetic perturbations involved in cancer or other common complex diseases. Already, extensive genetic studies using conventional resequencing techniques have identified genetic alterations for several common cancer types, such as breast, colorectal, brain and pancreatic cancers. Moreover, genome-wide association (GWA) studies now identify further novel loci and genetic variants. The next step will be to complete the Cancer Atlas, taking into consideration evidence from both resequencing and GWA studies. Many problems exist in completing the definitive causative catalog for each solid cancer. The next endeavor will be to understand the complex gene–environment interactions. This may clarify the cancer tumorigenesis puzzle. However, at present, neither has the genetic make-up been completed nor is there a standard approach for measuring the gene–environment network.
▪ How reliable is the information obtained from genome-wide association studies in terms of identifying the risk of cancer?
By using large sample sizes and replicating experiments, GWA studies have already provided a large number of novel susceptibility loci and SNPs through analysis on genotyping platforms with less than 500,000 SNPs. Future unbiased studies using whole-genome scans with more than 1 million SNPs will further expand the catalog of susceptibility loci.
However, GWA studies do not provide information on how these loci interact and which of these variants are causes or simple effects. Therefore, we cannot infer from GWA studies whether DNA variants are in fact responsible for phenotypic variation (cancer heterogeneity), and if so, which DNA variants are responsible, and how they influence phenotypic variation.
▪ What other factors can influence this risk?
The polygenic model is currently widely accepted. A large number of genetic alterations contribute to an increased risk of cancer, although we do not know how many contribute for each solid tumor. These genetic variants are either heritable through evolution or are caused by environmental factors or randomly. There is a universal consensus of the high complexity and heterogeneity of each solid tumor.
▪ Are there any advantages of copy number variant analysis over SNP analysis in the identification of these target regions? Once a potential target region has been identified, what steps would be taken in the prevention of that patient from developing cancer?
Indeed, deletions, duplications and rearrangements may play an important role in tumorigenesis. Companies now offer genotyping platforms with more than 1 million copy number variants (CNVs) for GWA studies. We should analyze new data to assess the extent to which CNVs are contributing to cancer risk. There is some estimation that approximately 80% of risk is attributable to SNPs and 20% to CNVs. Yet, no precise estimates can be made.
▪ In your opinion, what are the advantages and/or limitations of personalizing preventive interventions for cancer?
Personalized, risk stratification-based primary prevention is the greatest goal for the 21st Century. This means that, according to each individual’s risk of developing cancer, cardiovascular or other common diseases, the most appropriate intervention would be applied at the right time. A simple example is as follows: a woman with a very high risk of developing breast cancer would have undergone prophylactic surgery or intensified surveillance. By contrast, for another woman with a low risk of developing breast cancer but a high risk of cardiovascular disease agents, significant lifestyle changes such as giving up smoking, taking up physical exercise, diet alterations and other lifestyle modifications would be more beneficial. However, the major problem is whether or not this ideal thought is realistic. At present, it is unknown whether an individual’s risk of developing a common complex disease is in fact predictable and whether it can be translated into clinical practice in the future.
▪ What do you consider to be the most significant limitations in the technology of genomic microarray profiling for the identification of potential cancer biomarkers and how do you feel that these could be optimized?
Despite an initial over-enthusiasm a few years ago (including from myself) for the potential clinical implications of gene-expression profiling data, many scientists are now skeptical. My concern is that molecular profiling alone is unable to provide accurate estimates. Cancer is too complex a disease; functional data alone without studying DNA variation have less probability in predicting clinical outcomes (phenotype).
▪ Do you feel that genomic studies could be integrated with gene expression studies in order to provide a more global picture of a target region?
Yes. But all the genetic and genomic data collected by both conventional and third-generation sequencing technology should be analyzed. The interpretation of all this information, however, is complex. Current conventional standards in the prevention and treatment setting should be considered. Only based on the available conventional evidence can we build innovative cancer network models to predict the genotype–phenotype mapping.
▪ What are your thoughts on the ethical issues surrounding the identification of patients that are predisposed to cancer?
It is very important, although science is the truth and we should learn how to approach this truth. However, at the same time, strict regulations are required to protect individuals at higher risk of cancer than others. This information might introduce discrimination between the private insurance and public sector, competition in the workplace and too many other disadvantages.
▪ How do you see the field of personalized genomics evolving over the next few years?
Society, individuals and industry have all shown a major interest in personalized biomedicine. At the same time, many companies want to make money and offer genetic tools. However, the data generated by these companies are still scarce. The competition for more data will yield the development of more reliable, faster and cheaper next-generation sequencing machines, as well as an explosion in data collection by GWA studies and whole human genome scans. There are some calculations that over the next few years, the quick sequencing of an entire human genome will be possible for less than US$10,000 or $5000. We will then need large-scale, prospective, population-based studies to compare the two groups (patients and controls) using all available data from conventional and personal genomics research.
However, limitations also exist. For example, the pharmaceutical industry does not enhance a patient-tailored targeted therapy. Too much money has been invested into a single-gene-based development of targeted agents. They insist on the motto ‘one agent for all’. The cost of tumor genotyping and for research to understand the DNA changes responsible for response prediction is too high. Practically speaking, in order for this scientific endeavor to be achieved, the development of too many agents will be required. Ultimately, few patients will receive a specific targeted agent. In view of the current economic crisis, this approach appears to be less realistic. The path towards personalized medicine has been opened. However, it is too long and myriad problems still need to be overcome.
Financial & competing interests disclosure
The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.
Related Research Data
