The context for the thematic grouping of rare diseases to facilitate the establishment of European Reference Networks

A significant body of work has been conducted around RD ERNs and the likely benefits have been well documented [1, 5‐7]. ERNs for RD can be established considering different aspects, such as: disease frequency and prevalence, cost-effectiveness, or the need to reach sufficient numbers of patients to increase expertise and deliver the best quality of care. However, independently of the established priority, a RD ERN to be effective should be based on the common effort of different stakeholders and build on what is already present in the community. The principles for the establishment of RD ERNs were enshrined in the EUCERD recommendations on this topic in January 2013 [8].

The concept behind the implementation of ERNs as a model for healthcare provision has clearly evolved from the pure aim to improve cost-effectiveness to encompass improving quality of care and equity in access to healthcare. This shift makes it relevant to consider that the priority around RD ERNs should be reaching sufficient numbers of patients to increase expertise and deliver the best quality of care. Therefore, one of the main challenges for the society in general and for the rare diseases stakeholders in particular is to determine how ERNs should be organized i.e. how many should be created and how should different diseases be grouped in order to achieve the above-mentioned priority.

A network, in social sciences, is defined as an association of individuals sharing common interests and providing mutual support and information [9]. Networks have the potential to achieve a large coverage depending on the number of members involved. When we transpose these definitions to the organization of medical care, ERNs are an excellent model to overcome some of the specific problems of rare diseases: scarcity of patients, resources and expertise. ERNs with a broad scope will allow the rare disease community the possibility of reaching a larger number of patients and a more diverse range of rare diseases.

With respect to existing networks, ERN status will need to represent a clear added-value such as: easier adoption and spread of innovations in medical science and health technologies; facilitation of medical training; faster dissemination of agreed standards of care and general knowledge in the area of expertise; and increased profile and recognition of the participants. ERNs will attract the best experts and more patients and although they are supposed to facilitate cross-border health care it is expected that the pooling of resources and the e-Health solutions will decrease the burden of travelling for the patients. Another expected though not explicit benefit is that the Networks and their members will be in a better position to apply for research funding and further develop and accelerate basic and translational research.

Existing networks in the RD field have arisen “ad hoc” in response to specific funding calls either from DG Sanco (now Santé) or from DG research. The diseases covered by these networks are often of low/very low prevalence and the networks highly focussed. The aims of the networks are highly variable and their sustainability has been a major issue (Table 1). No mechanism has yet been in place to ensure that all or even a majority of RD patients might have access to a network for their particular disease. In the process of drafting and adopting the EUCERD recommendations on RD ERNs, this issue of inclusivity was debated in some detail. Patients’ organisations such as EURORDIS strongly advocated that ERNs should be inclusive and should not be created for every single RD, on the grounds that this would leave many patients without an ‘umbrella’ ERN [10]. Creating 6–8000 individual ERNs (as many as the estimated number of rare diseases) would be impossible, and not desirable from a clinical perspective.

Following the publication of the delegating and implementing acts, further discussion within the CEGRD indicated that more guidance was needed on the grouping of diseases into ‘families’ that can be addressed by comprehensive ERNs as a realistic and constructive approach.

In order to be successful, ERNs need to take into consideration existing national practices and networking systems which entails both opportunities and challenges. Existing networks and national healthcare authorities need to be made aware of the economical, scientific and patient care benefits of joint versus single disease networks.

The EUCERD Recommendations on Rare Diseases European Reference Networks (RD ERNS), published on 31 January 2013, aimed at provide guidance to the Member States and the EC on the criteria needed to be fulfilled to establish RD ERNs [8]. These recommendations were generated taking on board the results achieved by pilot ERNs funded through DG Sanco or DG Research between 2008 and 2013.

After the publication of the Delegated and Implementing Acts in March 2014 the EJA conducted a mapping exercise to explore the areas where the previous recommendations needed further attention from the EC Expert Group on Rare Diseases. That mapping exercise was the basis for an addendum to the EUCERD recommendations. Based on the content of the EUCERD Recommendations and the content of the Acts the Addendum to the Recommendations suggested an illustrative grouping of RD as a rational approach to RD ERN planning and to ensure coverage of all RD.

We based our exercise about the grouping of rare diseases on several possible models such as: the alignment of diseases by clinical area (e.g. Neurology, Neuromuscular, Psychiatry, Skin, Kidney, etc.), by clinical group (e.g. Genetic Disorders, Metabolic Disorders, Epilepsy, Oncology), clinical intervention area (e.g. Transplantation, Gene Therapy, Radiotherapy), shared molecular aetiology [11] (e.g. Underlying disease mechanism or pathway such as fibrosis or inflammation) or mixed models.

Several successful examples of centres of expertise and international networks for RD already exist, which may be viewed as concrete, trusted ‘solutions’ in their field of expertise and should be approached as case-studies from which one can derive important lessons for future ERNs. These tend to be grouped by clinical area or clinical grouping (e.g. rare anaemias, neuromuscular diseases, metabolic diseases, rare epilepsies, etc.).

An ERN based on a specific treatment area, such as for example Gene Therapy, although attractive when considering clinical trials or therapy development would inevitably leave numerous rare disease patients without a “home”. This model would obviously be involved with different clinical areas and could be regarded as transversal, cutting across numerous medical specialties. Such a transversal ERN, could not replace ERNs based on clinical areas or clinical groups and would have to co-exist alongside these.

ERNs based on shared molecular mechanisms would, as the former model, be a way of addressing problems related with clinical trials and drug development. Drugs that target a molecular pathway that is common to multiple diseases can, in principle, be used to treat more than one disease. However, though the concept is interesting in particularly from a research perspective, such an ERN might prove rather restrictive and too “single issue”, especially given the multidisciplinarity inherent in the expectations asked of an ERN.

We therefore compared five examples of disease grouping based on clinical areas/groups (the Classical Medical Ontology; the Orphanet classification; the EURORDIS preliminary proposal for grouping of rare diseases; the French filières; and the UK Research Model as established by the NIHR Translational Research Collaboration on RD). This comparison enabled us to propose a preliminary strategy for grouping rare diseases based on the merging of the common items. The classical medical ontology is the one used in Internal Medicine textbooks and is usually organised according to major organs and systems. Although incomplete for RD, it is widely used for teaching and is imbedded in the physicians’ way of thinking. A well-known example is the classification used in Harrison's Principles of Internal Medicine [12].

The Orphanet classification of rare diseases is structured around 116 groups of diseases. For the purposes of creating ERNs, 116 ‘categories’ would be excessive; however, in terms of thematic grouping Orphanet proposes 31 RD categories [http://​www.​orphadata.​org/​cgi-bin/​inc/​product3.​inc.​php]. It is a useful scientific classification; however, when dealing with ERNs there are some overlaps and some redundant groups. If, for instance, we consider the broad group of “Rare genetic diseases” should sickle cell anaemia be in the scope and expertise of an ERN for rare genetic diseases or for rare haematological diseases? Is it practical to have an ERN devoted to rare intoxications defined by Orphanet as rare intoxications due to medical products, or should these be incorporated in ERNs dealing with, for instance cardiac disorders for digitalis intoxication or with oncology for cytostatic intoxication?

The EURORDIS proposal, presented at the workshop “Rare Disease European Reference Networks (RD ERNs) and the use of structural funds to support activities for RD” held in Rome on the 28-29th October 2014 organises ERNs by clinical area and was based upon research undertaken by EURORDIS collaborators. This was an outline document, kindly made accessible to us by EURORDIS, which drafted groupings to facilitate discussion with the Council of National Alliances, Council of European Federations and EURORDIS members (Table 2). The Second French National Plan for Rare Diseases mandated the creation of ‘filières de santé maladies rares’. These “French national rare disease healthcare networks” aim to coordinate all the missions and activities of groups of centres of expertise and related competence centres in charge of coherent groups of related rare diseases. At present 23 official networks were created [http://​www.​sante.​gouv.​fr/​les-filieres-de-sante-maladies-rares.​html]. The ‘Filières de santé maladies rares’ were adequately based on what was already in place in the French health system. The main problem stays when we try to extrapolate these country-specific pathways to a broader coverage. Also, the ‘themes’ range from relatively broad (e.g. inborn metabolic diseases or developmental anomalies and malformations) to disease-specific (e.g. Amyotrophic Lateral Sclerosis). Given the lack of therapeutic options for the vast majority of the 6–8000 rare diseases, there is a necessary link between research and clinical care provision; therefore, we have also considered in our analysis the example of how RD have been grouped by the UK National Institute for Health Research (NIHR) Translational Research Collaboration in Rare Diseases (Table 3).

After analysing these different models we have tried to merge the common lines having in consideration that the grouping strategy has to incorporate the need of having healthcare providers with different roles/expertise in any network, the fact that a healthcare provider may take part in different ERNs, the need to have a significant collaboration across ERNs as well as within, and that fluidity and inter-communication between ERNs should be ingrained in the structure of the network. Another important question that was debated was whether a dedicated ERN for undiagnosed patients would be feasible or desirable. Here the distinction between truly undiagnosed patients (those in whom a precise diagnosis is not possible and who may require access to research programmes for a diagnosis to be achieved, such as is available via the NIH undiagnosed scheme in the USA) and the patients who are known to have a particular type of disease but not yet a precise diagnosis, needs to be made. As an alternative to a dedicated ‘undiagnosed’ RD ERN, it is suggested that each ERN should maintain a ‘forum’ for undiagnosed patients suspected to fall within their area of expertise. Considering what was stated about how RD ERNs should be structured, the necessity to encompass all rare disease patients and the merging of the different models we have studied, we suggest the constitution of 22 broad groups (Table 5). This list was posteriorly presented to the Commission Expert group on Rare Diseases for discussion. Some changes were suggested and the final grouping (Table 6) was published as part of the Addendum to the EUCERD recommendations of January 2013 [8].

As with the previously described models, the proposed model is not perfect. However some of the difficulties can be overcome as long as the ERNs are organized as flexible structures with capacity to adjust to the reality and maintain enough plasticity to share patients amongst them, in particular the undiagnosed ones (Fig. 1).