Biobanking residual tissues

Abstract

Health-care research relies largely on human materials stored in highly specialised biorepositories. Medical translational research on tissues can be performed using a variety of resources in distinct situations. The best known is the secondary use of pathology archives where paraffin-embedded tissues are stored for diagnostic reasons. Another is collecting and storing frozen material obtained from leftover surgical diagnosis. Such residual tissues can either be used directly in research projects or used in the context of a clinical trial with new interventional medicinal products. The latter can make the regulations governing the use of these materials for medical research much more complicated. The use of residual materials is very distinct from biobanking projects for which tissue is specifically collected. This article describes the consequences of using residual human material from different sources in distinct situations and why signed informed consent is not always the preferred choice of individual countries regarding the use of residual material. In addition, signed informed consent is overdone when using residual tissues in medical research. We maintain that the opt-out system is a balanced choice if certain requirements are met, relating to sufficient transparency about using residual tissue for research, the purpose of such research and to the confidentiality of the data used in that research. Finally, the international exchange of samples can be based on the laws and regulations of the countries of origin. Respecting these form the basis of what can and cannot be done in the country where the research on the samples is being performed.

Appendix

Text Box 1 Life cycle of frozen tissue samples

Text Box 2: Typology of data

There is considerable confusion about the terminology of coded, anonymous or identifiable tissue (Knoppers and Saginur 2005; Elger and Caplan 2006; van Veen et al. 2006; Nietfeld 2007). This confusion also applies to epidemiological data in general, without the context of research with tissue. Several proposals have been put forward (van Veen et al. 2006; Nietfeld 2007, ICH 2007). As tissue is always data as well—data are connected to the tissue and can be derived from the tissue—but not the other way around, the correct terminology following from data protection legislation should be applied. That terminology dates from before research with tissue entered the scene. The terminology should be neutral regarding the admissible use of data or tissue. Such assessment should come later, in view of the applicable privacy, human research protection legislation and the specific ethical assessment when research with tissue is concerned.

Here, we start with the typology of Directive 95/46/EC, which only recognises two kinds of data, namely personal data (falling within its ambit) and anonymous data (not covered by the Directive). Personal data those which are identifiable by the entity having those data, in the terms of the Directive ‘the controller’ or a ‘processor’ working on behalf on the former. Data can be directly or indirectly identifiable. Directly identifiable are data with direct identifiers attached to them, such as name and address. Indirectly identifiable data are those which are identifiable through the attributes attached to a data record, such as profession, age and zipcode.

The Directive states that the criterion whether data are re-identifiable is: ‘the means likely reasonably to be used’ to (re)identify (Recital 26). Obviously, such a criterion leaves room for discussion. The Directive has instituted a so-called ‘Working Party’ composed of members of the European National Data Protection Authorities. In 2007, the Working Party gave an Opinion on the concept of personal data (Working Party 2007). It mentions the high threshold, which must be used before data can be considered not indirectly identifiable. ‘Likely reasonably to be used’ becomes de facto ‘which theoretically could be used’. The Opinion also discusses the data chain and coding at some point in the chain. Coding at one point in the chain does not by itself influence whether data at a later point in the chain are anonymous or not. For that question, it only matters whether the entity holding the data at that later point can re-identify the data subject or not. The Opinion mentions the example of Case Record Forms (CRFs) which are sent to the sponsor of a clinical trial. The Opinion considers them anonymous at the level of the sponsor. CRFs are always two-way coded. Hence it is wrong—as is often assumed—to consider two-way coded data or tissue always to be personal data or tissue at the level of the receiver of the data or tissue, at least in terms of Directive 95/46/EC. But neither are they always anonymous data or tissue. It will depend on the level of security of the pseudonym or code number and again the re-identifiability of the data attached to that pseudonym. Only if those data are sufficiently aggregated (instead of birth date, year or 5-year class, etc.) they can, in combination with the aggregation level of the other data, be considered to be not re-identifiable.

Hence the following typology of data can be made

Anonymous data	Fully anonymous data
	Coded-anonymous (pseudonymised) data
Personal data	Indirectly identifiable data	Coded but either coding insufficiently secure or aggregation level too low
		Not coded, but aggregation level too low
	Directly identifiable data

In the text, we come back to the threshold for not considering data not indirectly identifiable. When we use the term ‘absolute anonymity’ we refer to a situation that data are anonymous by all standards instead of not by ‘means likely reasonably to be used’ as mentioned in the data protection Directive or where, in terms of the Sect. 164.514 under b.1 of the US Department of Health and Human Services ‘Privacy Rule’ under HIPAA, the risk for re-identification would not be ‘very small’, but zero.

It might be argued that this terminology is insufficiently nuanced for research with tissue. In that context it matters whether data accompanying the tissue at the research domain (the end of the data chain) are two-way coded or not. Only in the first case is individual feedback possible. If one considers this a relevant factor, one should make a further distinction in the second box on the right-hand side between two-way or reversible coded (at the source) and one-way or irreversible coded (at the source).