According the previous publications, independently of their clinical functions (which form the focus of most studies), MTB are also starting to coordinate the flow of biological samples, genetic data and information between patients and clinicians, but also between researchers at academic medical centers and industry, through clinical trials and retrospective studies (Fig. 3). If the intensity of these flows continues to increase, then, in the future, MTB or similar structures may play a much greater role in deciding who has access to these data. Thus, the MTB or similar structures may become major decision-makers at academic medical centers, facilitating the transfer of data from care to research contexts.
This would also improve the tracing of important decision-making processes, and their organization around a group of field experts in DM. In this era of major revolutions in science and technology (big data, data mining, machine learning, artificial intelligence, cloud computing, etc.), it has been suggested that the experts involved in these decision-making processes will no longer be exclusively medical doctors. Instead, these groups of experts will include scientists, statisticians and informaticians. These changes are likely to have a strong impact, and they raise several ethical issues, some of which are already of importance, including the reasons for communicating with patients and how best to do so, particularly as concerns the biobanks, databases, genomic or bioinformatic processes, clinical trials and retrospective studies developed through these flows. They also add to the risk of unequal access to these technologies on the basis of socioeconomic or genetic criteria.
Many clinical trials may be considered as treatment options because they provide some patients with a “last chance” of recovery. Nevertheless, clinical care and clinical research differ in that the objective of care is to be of benefit to a patient, whereas the objective of research is to gain new knowledge (Jefford and Moore 2008). The information provided to the patient when consent is sought also differs considerably between clinical care and research: for research, the information and the goals are explained in the conditional tense, because there are far fewer certainties than for clinical care. Indeed, one of the first objectives of clinical trials is to determine whether the new treatment works better than the old one. Furthermore, the new treatment may entail unknown or theoretical risks, which is less likely to be the case for a treatment that is already authorized. Thus, the first ethical issue raised by a MTB is the following: if a treatment is prescribed on inclusion in a clinical trial, can it still be considered to be clinical care?
This is a particularly important issue, given that scientific studies of a MTB have reported essentially inclusions in clinical trials rather than the prescription of authorized cancer treatments (Le Tourneau et al. 2012; Kamal et al. 2015; Sohal et al. 2016; Schwaederle et al. 2014). Furthermore, there are several different levels of clinical trials. For example, phase I clinical trials are designed to assess the toxicity of a new treatment in human subjects (acceptable dose levels) and are often conducted on patients presenting treatment failure (Brown et al. 2011). Therapeutic benefits are more likely to be seen in phase II and III trials, which test therapeutic efficacy rather than toxicity. However, the issue is more complex for targeted cancer therapy, and several studies have reported direct therapeutic benefit in phase I trials (Khan et al. 2016), even though the patient’s clinical condition (Sohal et al. 2016) and the quality or quantity of biopsy material can be a barrier to participation in such clinical trials (Lim et al. 2016).
For these reasons, “possible care”, “conditional care” or “potential care” should be considered, rather than just “care”, and these concepts should be explained in more detail to patients to ensure that they understand the consent form they are asked to sign. These adjectives qualifying “care” are therefore of great importance, because a short-term individual benefit is for the moment, merely possible; in theory, trials are seeking to obtain long-term collective benefit. But can this “care modify the “theory”? Perhaps, and it might even be a good thing if it did, because the frequent inclusion of cancer patients in clinical trials necessitates a new strategy for communication between patients, clinicians, researchers and industry. We need to make use of ethical discussions about the content and form of IC, to find new ways of improving the patients’ understanding of their participation in clinical trials (Kelley et al. 2015; Kao et al. 2016) (Fig. 3).
The publication of the first draft sequence of the human genome (Lander et al. 2001; Venter et al. 2001), and its completion by the Human Genome Project marked an increase in the strategic importance of DNA banking and data collection. In the past, biological samples were stored in a single laboratory, but large collections of DNA samples are becoming increasingly common in human genetics (Thornton et al. 2005). DNA can be obtained from a number of potential sources, including the blood, the cell and tissue banks of hospitals and academic research centers, and it has been estimated that there are already several hundred million biological samples stored in such repositories (Swede et al. 2007).
Biobanks are becoming increasingly important for the establishment of research infrastructures (Castillo-Pelayo et al. 2015), and the systematization of NGS, clinical trials and retrospective studies in cancer may render these repositories of even greater importance, through a cyclic phenomenon. As more clinical trials are set up and more patients have their DNA sequenced, there will be more biological samples, genetic data and clinical information stored in biobanks, leading to more retrospective studies being carried out and more clinical trials being designed.
Retrospective studies could play a key role in completing the loop. However, researchers cannot know in advance which topics they will address through retrospective research studies. It is not, therefore, possible to provide patients with “perfect information”, that is to say, a description of all possible research purposes in advance, when they sign an informed consent form. Institutional review boards and ethics committees generally require consent to disclose all of the ways in which samples and data may be used and to ensure an acceptable balance between risks and benefits (Godard et al. 2003; Rothstein 2002). This is why some recent studies have tried to determine whether patients prefer a “re-consent” or a “broad-consent” process, but no clear consensus has yet to emerge (Edwards et al. 2016; Goodman et al. 2016).
The increasing number of retrospective studies being carried out on cancer, using data collected in clinical studies, requires a new strategy for communication between patients, clinicians, researchers and industry, and ethical reflections are required about the content and nature of the informed consent collected, with the possibility of returning to patients if necessary (Edwards et al. 2016). This approach should improve the patients’ understanding of research protocols, increasing their autonomy and their trust in academic medical centers (Fig. 3). However, it will require a particularly well-adapted form of informed consent.
Informed consent is much more than a simple form of communication between patients, clinicians, researchers and industry (Annas 2001). With the increasing numbers of retrospective studies based on data from clinical trials in the field of cancer, we need to consider the form of informed consent given, and not just its content, at all academic medical centers at which there are close ties between care and research. Recent innovations in information technology (IT) could drive changes in strategy concerning communication between patients, clinicians, researchers and industry (Kaye et al. 2015).
Trust is based primarily on communication. If patients feel that they can communicate freely, anywhere, at any time and in complete safety, they are more likely to trust the system and to be willing to share their data and biological samples. An electronic informed consent (e-IC), or dynamic consent (Kaye et al. 2015; Steinsbekk et al. 2013; Dixon et al. 2014; Williams et al. 2015; Spencer et al. 2016) system would open up new possibilities: (1) for patients to determine whether or not to give consent, at the time and place of their choosing, (2) for clinicians to ensure, in real time, that a particular piece of information has been transmitted to a particular patient and to determine whether or not the patient concerned has given consent and (3) for researchers and industry, to facilitate the use, in real time, of particular samples or data (Karlson et al. 2016).
The FDA [according to a draft document available from the FDA website (www.fda.gov)] has identified IT as a good way of improving communication between patients and clinicians through multiple media: text, graphics, audio, video, podcasts and interactive websites, biological recognition devices and reader cards to relay information about the study and to obtain informed consent. The FDA draft discusses the possibility of e-IC. Nevertheless, the FDA draft stresses that the content requirements for e-IC should be identical to those for written informed consent: the information should describe and explain clearly the purpose of the study or analysis and it should be easy for the patient to understand the aim of the research.
In general, e-IC facilitates more interactive and dynamic exchanges of information and improves the traceability of biological samples, genetic data and information (Kaye et al. 2015). This process has already been well understood by a number of companies in the US, including Google, Amazon, Facebook, Apple (GAFA), Microsoft, 23andMe, and Helix, a new company that will soon begin trading.
A global view is required to visualize how the rapid sharp decrease in the cost of DNA sequencing and cloud computing has initiated a flow of biological samples and genetic data directly between patients, researchers and industry, leaving some academic medical centers excluded (Fig. 4).
The situation on the ground has changed radically since the 1990s, and several American companies specializing in genomics or informatics appear to have established two-sided market (TSM) platforms through the constitution of biobanks or databases with two kinds of customers: individuals seeking information about their own genes, with or without the need for a prescription from their physicians, and industrial and researcher clients seeking access to the genetic, web behavior and self-reported data obtained from large numbers of people through a commercial contract involving dynamic, electronic informed consent.
23andMe is probably the best known of these companies. Since 2006, this genetic testing company has collected biological samples, genetic data and information from 1,000,000 individuals, at least officially through a direct-to-consumer (DTC) online genetic testing service providing a genetic ancestry report and a genetic health report. In reality, the primary objective of the company was not the provision of DTC testing services, but the establishment of a TSM: promoting itself as providing predictive testing for human genetic diseases and ancestry at a low cost to consumers, whilst establishing a high-value database/biobank for research and industry, for scientific and financial gain (Stoekle et al. 2016). The commercial agreements between 23andMe and Genentech for Parkinson’s disease provide strong evidence of the commercialization of the company’s biobank and database (Mullard 2015).
A new company of the same kind, Helix, seems to be emerging in San Francisco, as reported by the MIT Technology Review. This company will collect sputum samples from customers (patients or healthy individuals), who will buy a DNA application for their smartphones. The company will then sequence the exomes of its customers, thereby generating and storing large amounts of genetic information that it will make available to consumers. The entry price for customers is low, at about 100 dollars. This company is based on a TSM model, but it is also a “pay-as-you-go” model (Regalado 2016a). This brings us back to fundamental questions about the content and form of informed consent, but also to questions about the ownership of biological samples and genetic data. Genetic data have become a commodity like any other, at least in North America and some European countries (UK) (Mullard 2015; Dorfman 2013).
However, the ownership of genetic data is not a straightforward issue. First, from an economic point of view, these data must be considered “public goods” (Stiglitz 1999), that is, goods that are both non-rival and non-excludable (Levêque and Ménière 2003). They are non-rival goods because they are not exhaustible: the consumption of public goods by one individual does not prevent another individual from using or consuming the same public goods. They are non-excludable because it is difficult to prevent a particular individual from using them. Once disclosed, genetic data can circulate freely and, in the absence of specific legal provisions, can be used by all, even those not participating in the sequencing or analysis processes.
As a direct consequence of these characteristics, genetic data are not considered to be “scarce” and their ownership cannot be justified on an economic basis, particularly at a time at which the price of sequencing is rapidly decreasing.
On the contrary, it would be appropriate to ensure open access to these data, to prevent their use for research and care being limited to those controlling them. Even if these data cannot be appropriated as such, economic actors try to protect them with legal devices, such as exclusivity agreements, confidentiality agreements or intellectual property rights (Stiglitz 2007–2008), such as those used to protect databases in Europe. Such protective arrangements provide the holders of the data with the power to control downstream markets in care and research (Kitch 1980). This control raises serious ethical problems if it jeopardizes access to care at a reasonable cost for all, or favors a move from social insurance systems based on solidarity to systems based on individual ability to pay.
Other questions are also emerging about the consequences of increases in the flow of genetic data between patients and industry: will industry always need academic medical centers for the development and testing of new drugs and tools? Will patients always need academic medical centers, or even physicians, for diagnosis and treatment? These questions are clearly relevant to the large numbers of individuals subscribing to offers of this kind [>1,000,000 over only 10 years for 23andMe (Stoekle et al. 2016)] and likely to request the new blood test for cancer being developed by Illumina. Indeed, the company recently said that it will form a new company (Grail) to develop a blood test, costing $1000 (or less), for detecting cancers before the onset of symptoms. Illumina predicts that this blood test will reach the market by 2019 (Regalado 2016b).
The Pan Cancer Analysis of Whole Genomes project has demonstrated that the use of cloud computing is much faster and cheaper than the use of conventional academic data centers for the analysis of large amounts of data. In this context, another approach to ensuring free access might be to ensure that all major genetic data be uploaded onto academic and commercial clouds, to prevent private companies, such as GAFA, from obtaining full control over these data (Stein et al. 2015). This process could be managed by MTB and e-IC, reducing costs by keeping the patients out of hospital but staying in real-time communication with them. IT could be used to manage all communication facilities for the development of cancer research, but it would not be competent to manage the message. We believe that academic medical centers remain the best structure for this purpose. This principle seems to have been applied by the Geisinger Health Institute (US) since 2006 [Geisinger website (www.geisinger.org)]. However, the final decision will probably depend on whether societies choose to organize their social insurance systems on the basis of solidarity, as in France, or according to individual ability to pay, as in the United States and many other countries.