Oxagen is a Clinical Genomics Company investigating the genetic basis of common human disease to discover new diagnostics and better targeted therapies. Founded in 1997 as a spin-out from the University of Oxford, the Company now has 75 employees, and is based on Milton Park near Abingdon, Oxon (see Attachment 1 for more details).

What current projects involve collecting genetic information on people in the UK?

  We are running nine research programmes searching for genes in which variants can lead to increased susceptibility to human disease. These are all common diseases, which have a complex underlying aetiology involving multiple genetic determinants, a large environmental component and chance. The diseases we are studying are: heart disease, asthma, inflammatory bowel disease (Crohn's disease and ulcerative colitis), osteoporosis, endometriosis, type two diabetes, polycystic ovary syndrome, autoimmune thyroid disease and psoriasis.

  The projects rely on assembling large collections of families, typically with two or more affected siblings. These are identified by our clinical collaborators through clinics or retrospectively by examination of their own databases. Blood is collected (typically three x 10 ml), and we prepare DNA for genetic analysis. The DNA is genotyped for 400 genetic markers—chromosomal "sign posts", and the data is analysed to reveal trends of co-inheritance between particular markers and disease. This narrows the search for the genes involved to a small number of chromosomal regions, typically around 30 million base pairs in length (about 1 per cent of the genome). These regions are then refined by typing the same and additional samples using more densely spaced markers to identify smaller regions of "association". The goal here is to identify DNA sequences that are more prevalent in affected individuals than in the general population. Detecting such sequences narrows the location of the disease gene to less than 300,000 base pairs. The final step is to identify all the genes and common variations in DNA sequence in the region, and to identify those sequence variations that affect gene function. It is only after this final stage that one can claim to have identified a "disease gene".

  All of this information—on pedigrees, clinical status and genotype—is kept on secure databases within the Company. Each sample is identified only through a reference number and the Company has no way of matching individual test results with the identity or address of a participant. Also, Oxagen's collaborators only provide clinical data relevant to the disease under investigation—the Company does not receive a copy of the patient's medical notes and can only gain access to them under medical supervision for verification purposes.

What other projects are about to start?

  We are planning to extend our autoimmune and inflammation programmes to examine arthritis. We are also considering participation in a European initiative to study the genetics of lung cancer susceptibility.

  We are interested in the idea of collecting a very large (around 1 million) prospective population sample that would be used to validate the diagnostic potential of genetic tests for disease susceptibility, and to examine gene-gene and gene environment interactions. The health of individuals would be followed over a long period of time, allowing medical events to be correlated with particular genotypes. This would be a huge project, necessarily involving a number of interested parties, including the NHS and the major sources of medical research funding such as the MRC and the Wellcome Trust. The idea of such a study has already been floated publicly by others.

Are there collections of material (eg tissue samples) that could be used to generate databases of DNA profiles?

  This is not an area we are involved in. Our feeling is that existing tissue banks are not particularly useful in the study of disease genetics because the DNA would be of erratic quality and quantity, and there might be insufficient data on the source of the sample. An important exception to this is in the field of cancer research, where archived tissue samples are very suitable for the study of the somatic mutation events that occur during tumourogenesis.

Why are these genetic databases being assembled?

  All commentators agree that the completion of the human genome sequence is just the start of a massive enterprise to catalogue and ascribe functions to all the genes. Beyond this, the challenge is to understand how small sequence differences in the genes (known as single nucleotide polymorphisms or SNPs) influence gene function and hence disease susceptibility. We are well placed in the UK to capture much of the commercial opportunity that will flow from this next phase, in particular, the study of genetic diversity. Some of this will be carried out in computers, but hypotheses generated in silico will need to be verified in vitro and in vivo using increasingly sophisticated model systems—the whole panoply of functional genomics. Our belief is that gene function will eventually have to be studied in man, and that the study of how genetic variation influences disease susceptibility and progression is the ultimate in functional genomics.

  Genetic databases are essential if we are to piece together the complex jigsaw of common disease susceptibility. Our goal is to gain new insights into the molecular mechanisms underlying disease on which radical new treatments can be based. Other benefits will flow from our ability to classify disease more precisely—disease stratification—and to identify those most at risk from disease allowing us to target life-style advice, screening and preventative intervention.

How are these activities funded?

  We largely fund our own programmes, though a number receive an element of governmental support through grants such as the LINK scheme. A number of projects start as shared risk initiatives with academics, in which we contribute genotyping resources in return for an option to take the study forward should the results look encouraging.

What practical considerations will constrain developments?

  Our ability to assemble the large clinical cohorts, including affected and unaffected individuals, is essential if we are to identify the genes underlying common disease. This is already a costly and time-consuming activity, and the impact of any additional safeguards on the use of patient data would have to be carefully assessed to ensure that it did not render large-scale genetic and epidemiological studies impractical. The UK has the opportunity to be a world-leading centre for such studies. Of particular concern to us would be restrictions on the use by clinicians of hospital records and databases to identify potential study participants—or a demand for individuals to have access to the test results.

Are there alternative ways of fulfilling the objectives?

  Genetics is probably the most powerful tool for studying the operation of complex biological networks. This was shown originally for bacteria and viruses, but can now be applied to more complex organisms ranging from yeast through the nematode and fruit fly to vertebrates such as mice. With man, no one would want to carry out germ-line modification to test genetic theories. However, we can still study genetics in man by a study of existing variation. This ranges from the study of rare single gene disorders—which are essentially a human equivalent of knock-out mice (transgenic mice engineered to lack a particular gene), to the common diseases that are the major causes of human morbidity and that touch all our lives.

  Any study of human genetics inevitably calls for a correlation of genotype with phenotype (ie the effect on the individual) on a large scale, and there is therefore no realistic alternative to the use of genetic databases in some form.

What is the genetic information that is being collected?

  This ranges from simple statements of a family history of disease, through large assemblies of anonymous genetic marker information (DNA signposts with no impact on disease such as microsatellite sequences or single nucleotide sequences in intergenic regions) to specific genotypes which are known to affect gene function or influence disease susceptibility. Of course what is thought of as an anonymous marker might at some time in the future turn out to be an important determinant of disease. However, the overwhelming majority of genetic variation examined will be essentially neutral in effect, and will turn out to have no bearing on disease. One important ethical issue is what to do with information, initially thought to be of no significance, that subsequently turns out to have major implications for an individual or family. We view this as a judgement that we should not try to make, rather we would pass the data to the relevant supervisory clinician and let them make the judgement.

  Oxagen has developed and published its policies in respect of ethics and sample collection procedures (see Attachments 2 and 3). [Not printed]

How is it being stored and protected?

  In our Company, the genotype and phenotype data are collected into centralised databases on a secure server with access limited to certain Company employees. Collaborating clinicians and commercial partners are not given access to company systems directly or indirectly (eg via a VPN). The Company operates to a high standard of IT security to prevent illicit access to data from internal or external sources and is isolated from the Internet by a firewall device. The Company also prohibits the use of laptop computers for analysis and management of clinical datasets to reduce the risk of exposure due to theft. However, the ultimate guarantee against misuse of such information is that we do not have access to any patient names or addresses—individuals are tracked with internally generated unique identifiers. We do not operate irreversible anonymisation of the data, however, as we need to provide for re-bleeds and clarification of inconsistencies in phenotype data. This is allowed for by sending the relevant unique identifier to the collaborating clinician.

How do the organisations involved see their responsibilities regarding privacy, consent, future use, public accountability and intellectual property rights?

  We take our responsibilities with regard to privacy extremely seriously. All our studies involve fully informed consent, with the use of the samples restricted to the field defined in the consent form (see Attachment 4 for an example from our thyroid disease programme). The clinical materials and data are collected through collaborating clinicians who act as guarantors of the patients' interests. An additional important principle is that we do not seek to own the patient samples, rather we leave ownership in the hands of the institution where they were collected. Our agreements provide for an exclusive period of commercial access to the samples which is typically three to five years. The samples therefore remain in the public domain, with access to other academic researchers controlled by a Research Steering Committee.

  We do seek to retain the intellectual property rights in discoveries made through the use of the samples, though our agreements provide for a percentage of any returns, including milestone payments and royalties, payable to the collaborating institutions. None or our agreements provide for payments to individual volunteers (other than expenses).

How do they see their activities in the area of genetic databases developing in the future?

  The scale of our genetic databases will grow exponentially, with the possibility emerging of selling access to the database to allow genotype/phenotype correlation to be carried out in the computer. This would be restricted to a subset of individuals who had been suitably consented.

What advances in sequencing, screening and database technology are they anticipating?

  Our ability to identify genetic variation is dramatically outpacing our ability to assign function to genes and SNPs. We are already in a position where there are in excess of two million common known polymorphisms due to the efforts of the SNP Consortium and Celera Corporation. The next wave of technological developments will allow us to genotype many thousands if not all of these, potentially across thousands of selected individuals. This will generate vast quantities of data and present huge problems in terms of analysis. The main impact of the technology will come from selecting a much smaller subset of functional and disease associated SNPs, and typing these in a more targeted fashion. Genotyping individuals for a set of these important SNPs will allow us to assign them to groups with elevated risk of certain diseases, allowing preventative measures to be taken and/or frequent screening to be used to capture disease in its early, more treatable stages.

What lessons should be learnt from genetic database initiatives in other countries?

  The initiative taken by Decode in Iceland leads us to the conclusion that whilst the large-scale database approach is feasible, the sensitivity surrounding a perceived monopoly of access means that it would have to be implemented as a public/private partnership in the UK.

Other points:

  All of our proposals for studies are subject to rigorous examination by the relevant local and multi-centre committees. We would welcome an initiative to provide a set of guidelines to ensure consistency of approach between committees at different centres.

  The wider use and standardisation of databases in all areas of the NHS would greatly facilitate genetic research. Ultimately we see great benefit accruing from the computerisation of all medical records. Access to such records by commercial concerns should be possible under suitable safeguards such as restrictions on use and anonymisation of data.

  The attitude of the insurance industry is critical in the development of genetic research. Creative incentives to the positive use of genetic and other risk factor data need to be found. The spectre of genetic discrimination needs to be avoided at all costs, as it could undermine the basis of shared endeavour between the research community, patients and their families.

  It is important to realise that a fundamental distinction can be drawn between genetic mutations that are deterministic in terms of disease outcome (single gene disorders such as thalassemia, muscular dystrophy and cystic fibrosis), and the polymorphisms that influence susceptibility to common ailments such as asthma and heart disease. Whilst the former allow us to make accurate predictions of individual risk, the common disease variations do not. Thus an individual carrying a risk allele for Alzheimer's disease might belong to a group with an increased risk of developing the disease, but themselves be at low risk because of other as yet unidentified genetic factors. There is therefore limited utility in genetic databases other than in a research setting.

Attachments: [Not printed]

  1.  Questions and Answers on genomics and Oxagen.
  2.  Oxagen's policy on sample collection.
  3.  Oxagen's policy on the ethics of genetic testing.
  4.  Sample informed Consent form.

Dr Mark Edwards

4 October 2000