Pharmacogenetics
Privacy and confidentiality
3.31 The implications for patients of DNA samples being used in research will differ depending on how easily their samples can be traced back to them, and whether the research is likely to give rise to information that may be of personal clinical relevance. Two related questions arise: (i) what level of anonymisation of samples is appropriate, and (ii) should individual patients be given feedback regarding tests carried out on their samples?
3.32 A recent Position Paper on Terminology in Pharmacogenetics by the EMEA/CPMP, published in December 2001, is a significant step toward the preparation of specific regulation on pharmacogenetics.27 The paper emphasises the urgent need for harmonised terminology for protocols and guidelines including pharmacogenetic testing.28 This is seen as a necessary requirement for the harmonisation of pharmacogenetic protocols in clinical trials. The paper sets out a classification scheme with respect to research samples used in clinical trials (see Box 3.2).
| Sample labeling code | Link between subject and data? | Records identifiable for clinical monitoring |
| Identified | Yes, directly | Yes |
| Single-coded | Indirectly via a code key | Yes, via protocol specified procedures |
| Doublecoded | Indirectly, via two code keys | Yes, via protocol specified procedures |
| Anonymised | No, key identifying the link between genetic data and the identity of the subject is deleted | No |
| Anonymous | No | No |
| Withdrawal of consent | Return of individual results? | Scope of subject’s privacy protection |
| Sample can be withdrawn with immediate effect for any prospective use | Possible | Similar to general healthcare confidentiality |
| Sample can be withdrawn with immediate effect for any prospective use | Possible | Standard for clinical research Conforms to principles of good clinical practice |
| Sample can be withdrawn with immediate effect for any prospective use | Possible | Standard for clinical research Conforms to principles of good clinical practice Offers added privacy protection over single code |
| Sample and data are not identifiable Sample cannot be withdrawn once key is deleted | Not possible | Data not linked to individuals |
| None | Not possible | complete |
3.33 Thus, identified samples are treated in the same way as samples acquired in other areas of medical practice. These samples are labelled with personal identifiers such as the donor’s name or social security number. This allows for easy retrieval of the sample from a study, should the donor wish to withdraw, and it is similarly easy to provide feedback. In the case of coded samples, there are ‘keys’ connecting the sample to the participant. The clinical investigator holds the key which links the patient’s name to the first code. In single-coded systems, the genetic researcher has access to this code. In double-coded systems, the genetic researcher only has access to a second code and the key linking the two codes together. Anonymised samples are like double-coded samples except that the identifying key is destroyed after the genetic and clinical information has been obtained. In the case of anonymous samples, no direct link between the sample and the donor exists from the time the sample is collected. Such a sample may be labelled with population information that indicates that the donor suffered from a particular disease, but it contains no individual identifying data. Of course, in principle, it would be possible to establish a link between an anonymous sample (or indeed, an anonymised sample) and the individual from whom it was obtained by matching the sample in question to another one from the same person. To do this, a second sample would have to be obtained, with the consent of the individual, and compared to all the samples in the database.
3.34 What are the relative merits of each approach? It is important to realise that pharmacogenetic analysis of samples could take place at various stages in a clinical trial. There might be basic laboratory research undertaken to examine how medicines interact with particular enzymes. Analysis could also be undertaken during a clinical trial to examine efficacy and safety in relation to genetic variation. Finally, once a medicine had been licensed, there might be additional research conducted if any problems were to arise in patients taking the medicine, although often at this stage new samples would be collected instead. There may be reasons to impose different degrees of anonymity in different types of research.
3.35 In general, anonymous samples are of little value in many types of pharmacogenetic research, since they do not allow the collection of data that links a particular patient’s genotype to his or her response to the medicine being studied. Knowledge of a DNA sequence is of no worth without knowledge of what happened to that patient when he or she took the medicine being studied. Anonymised samples have the disadvantage that they preclude the follow-up or monitoring of an individual’s condition in the longer term, once the key linking the sample to the patient has been destroyed. However, clinical information can be collected and linked to the sample prior to its anonymisation, which can provide sufficient information within the context of a clinical trial. Subsequent research in the postmarketing surveillance phase of a trial can be conducted on new samples in most cases.
Single-coded or double-coded samples are more likely to be appropriate if researchers wish to trace participants once a clinical trial has ended and also when the trial is taking place, since clinical information may be collected at various stages. This approach has the potential but speculative benefit for participants that if the research subsequently reveals information of relevance to the participant’s health or medication, the patient can be informed. However, if important information is discovered and the samples have been anonymised, participants may be re-tested as part of their ongoing clinical treatment (see paragraphs 3.44 – 3.49 for a discussion of providing individual feedback). Identified samples are rarely used. It should be noted that samples which are coded or identified could also later be anonymised and used in other research projects, depending on the nature of the consent obtained from participants.
3.36 In the specific case of pharmacogenetic research, we take the view that it is generally possible to obtain genetic and clinical information about a patient during a clinical trial and then to anonymise the samples. In most cases, new samples can be taken from patients suffering adverse reactions and from controls for the purposes of post-marketing surveillance without compromising the quality of the research. In some cases, for example, trials that last for a very long period of time, anonymisation would not be able to take place without compromising the goals of the research. There may also be auditing requirements imposed by regulators which entail that samples cannot be anonymised, even for a number of years following the completion of a clinical trial. We consider that to protect the privacy of participants in research, the greatest degree of anonymity should be imposed on samples, compatible with fulfilling the objectives of the research. Researchers should explain to prospective participants the implications of the manner in which samples will be stored for that participant.
3.37 It might be suggested that storing samples in a coded or identified form is acceptable only where the participant is aware of the precise nature of the research to be conducted. This could be argued on the grounds that future use of the sample could reveal information about the participant which he or she would not have wished to know. It can also be argued that, whether samples are anonymised or not, there should be limits to the use to which they can be put, since there may be some types of research to which the participant does not wish to contribute. Thus, a distinction is often drawn between ‘broad’ and ‘narrow’ consent. The latter refers to instances where a sample is only to be used for a restricted range of purposes, perhaps only for a single research project, or research in relation to one particular medicine or condition. Broad consent entails that patients agree that their sample may be used for a variety of future studies which it may not be possible to specify in any detail at the time of consent. Usually, but not always, these future studies will be within the same broad areas of research as the initial project. For example, some researchers may wish to use samples taken for pharmacogenetic research in general studies examining the genetic basis of disease. In practice, there is no dividing line between broad and narrow consent. The breadth of the research proposed could range from any biomedical research to a particular study.
Footnotes27 EMEA CPMP (2002) Terminology in Pharmacogenetics – Final Position Paper. Available: http://www.emea.eu.int/pdfs/human/ press/pp/307001en.pdf. Accessed on: 5 Mar 2003.
28 The system set out by the EMEA is similar to that proposed by the Pharmacogenetics Working Group, which distinguished identified, coded, de-identified (double-coded), anonymised and anonymous categories. Spear BB et al. (2001) Terminology for sample collection in clinical genetic studies, Pharmacogenomics J 1: 101-3. Different terms are used by the Medical Research Council in its publications. We use the EMEA terminology in this Report. The corresponding terms used by the MRC are as follows: linked (coded); linked anonymised (double-coded); unlinked anonymised (anonymised).