A scoping study to explore the cost-effectiveness of next-generation sequencing compared with traditional genetic testing for the diagnosis of learning disabilities in children

Definition of patient group

Valuing People,6 the 2001 White Paper on the health and social care of people with LDs, includes the following definition (p. 14):

Learning disability includes the presence of:

• a significantly reduced ability to understand new or complex information, to learn new skills (impaired intelligence), with;

• a reduced ability to cope independently (impaired social functioning);

• which started before adulthood, with a lasting effect on development.

This definition is broadly consistent with that used in the current version of the World Health Organization’s International Classification of Diseases, 10th Revision (ICD-10), although the ICD-10 uses the term ‘mental retardation’ (MR). 7,8 International Classification of Diseases, 11th Revision is in development (due by 2017) and it has been suggested that this publication is likely to use the term ‘ID’ to replace both MR and LD.

Mir et al. 9 explain that the term ‘learning difficulties’ is the preferred term amongst user organisations and disability writers, whereas ‘learning disabilities’ is generally used by service organisations. In this scoping study the abbreviation LD is used throughout to denote ‘learning disability’.

However, to add further ambiguity, a number of other terms included in published papers and other documents are used synonymously with, or include the definition of, LD. These include ‘mental handicap’, ‘ID’ and ‘developmental delay’ (DD). The term ‘DD’ is sometimes applied to younger children for whom a diagnosis of LD or MR cannot be established because the child is not yet old enough for a reliable cognitive assessment to be performed. However, the term ‘DD’ can be criticised for implying that ‘developmental catch-up’ might, or will, occur; for the vast majority of children who are tested this will not happen as they have a permanent LD. Some people prefer to use the term ‘DD’ for ‘developmental disability’ whereas others favour ‘disordered development’ (i.e. following a different developmental pathway) or ‘early development impairment’ (for children under the age of 4 years for whom accurate assessment of the degree of LD is very difficult). Sometimes the term ‘learning and developmental disability’ (LDD) is used to encompass both DD in younger children and LD in older individuals (e.g. by Newman et al. 10). However, in UK education circles, LDD specifically means ‘learning difficulties and disabilities’ rather than ‘learning and developmental disability’. Other terms used in the published literature include ‘global developmental delay’ (GDD), ‘cognitive impairment’ and ‘idiopathic learning disability’ (ILD), with ILDs being LDs with no known cause. An author from the USA11 describes ‘development disabilities’ as including ‘congenital malformations, cognitive impairment, and behavioural abnormalities’ (p. 121).

Different terminology is used for similar conditions in different countries, at different times and/or in different situations. For example, ‘mental handicap’ was widely used in Britain but is now considered to have inappropriate and negative connotations. MR is considered offensive in Britain but not in the USA, although it has now been replaced by ID.

Sometimes the terms ‘autism spectrum disorder’ (ASD) and ‘multiple congenital anomalies’ (MCA) are used when researching and discussing LDs. The inclusion of children with ASD in a study of a particular technology for diagnosing LDs may have a significant effect on its identified diagnostic yield because about 50% of people with autism also have an ID. Interest in, and research on, ASD has increased in recent years. Temporal comparisons of diagnostic yields for LDs from different technologies may be compromised if the cohort of children undergoing testing also includes a significant proportion of children with ASD (although the boundaries between LD and ASD are not clear-cut).

When discussing the findings of published papers, this report uses the terminology chosen by the authors of the paper (even if it may sometimes be considered offensive by some readers of this report). Elsewhere in this report, the terms LD and DD are used to encompass learning disabilities (as defined in Valuing People6) and to capture DDs developmental impairment in younger children respectively.

Prevalence

Various attempts have been made to determine the number of children with disabilities, including LDs; however, these have had limited success. 12,13 The report of the independent inquiry into access to health care for people with learning disabilities, Healthcare for All,14 suggests that around 2.5% of the UK population has a LD. Emerson and Hatton15 used two different approaches to estimate prevalence in England. The first involved extrapolating from a subset of locally held LD ‘registers’ (the estimated administrative prevalence) whereas the second made adjustments to the ‘register’ figures, which took into account the fact that these figures are likely to include only those with severe LDs (estimated true prevalence). Prevalence figures by age group are presented in Table 1 and estimated numbers of children with LDs, calculated using Office for National Statistics subnational population projections for 2012,16 are displayed in Table 2.

Diagnosis of learning disability

The first step in the diagnostic pathway occurs after someone (perhaps a professional, relative or friend) believes that a child may have a disabling condition or additional health needs. This concern may arise as part of a formal screening or surveillance programme. For example, it could occur pre birth if, for example, amniocentesis has shown possible Down syndrome; at birth for a condition with physical abnormalities; later during a consultation with a clinician for an unrelated purpose; or, in the case of less profound DD, when the child starts school.

A paediatrician generally carries out the assessment of LDs. The paediatrician will take a medical history as well as record physical measurements and assess the child’s developmental progress. Other health professionals may also be involved in the assessment, for example a child may be referred for a vision or hearing assessment or to a neurologist. 17

If the paediatrician suspects that the child’s condition may be linked to a genetic abnormality the child will be referred to a regionally based NHS clinical genetic service. The genetic service can use a number of diagnostic tools, including:18

• family history taking

• information gathering

• medical history and examination

• clinical investigation

• genetic testing.

Results from a genetic test can generally be obtained within 4 months. Some NHS centres offer a turnaround time of 6 weeks for specific tests. Certain specialised tests can only be carried out abroad.

The clinical genetic service provides genetic counselling. The extent of counselling depends on the result of the genetic test. Sometimes it is appropriate to counsel parents only, but in other cases the wider family may need counselling and, perhaps, also testing. For example, if both parents are found to be carriers (i.e. each parent has one ‘normal’ and one defective gene) then siblings could also be silent carriers of the condition. Even if the siblings do not have the condition themselves, they could potentially pass the mutated gene to their children.

The value of a diagnosis

In 1995 the American College of Medical Genetics [which changed its name to the American College of Medical Genetics and Genomics (ACMG) in 2011] sponsored a conference in Minnesota, USA. The purpose of the conference was to use available literature and expert opinion to produce consensus recommendations on the evaluation of MR. The benefits of a genetic evaluation, as identified at the conference, are displayed in Table 3. 19

In the UK, as part of a larger study looking at the implications of WGS for health, Gogarty20 used feedback from interviews (38 participants), a forum (46 registered members) and discussion groups (36 people) to develop recommendations for service change with regard to the clinical investigation of children with DD. Participants included parents, patients, carers and representatives from voluntary organisations. As part of the consultation exercise, parents were asked to describe some of the advantages and disadvantages of genetic testing for LD. Some of the most common responses are displayed in Table 4.

Ongoing research

There are currently three large-scale genetic research studies under way in England that are of particular relevance to children with LDs. All three are based in Cambridge.

• the Genetics of Learning Disability (GOLD) study21 was established in 2001 and aims to identify the genes on the X chromosome that contribute to significant ID and to lead the way to a greater understanding of the mechanisms by which ID occurs

• the Deciphering Developmental Disorders (DDD) study22 aims to advance clinical genetic practice for children with developmental disorders by the systematic application of the latest microarray and sequencing methods, to see whether these technologies can help doctors understand why patients get developmental disorders, and also to address the new ethical challenges raised

• the Specialist Pathology Evaluating Exomes in Diagnosis (SPEED) project23 is a NIHR BioResource – Rare Diseases initiative. One of its immediate aims is to develop more affordable DNA-based testing of rare diseases for which the gene is known.

Traditional diagnostic methods

The advent of next-generation sequencing technologies

Data quality

A recent Briefing Note published in 2014 by the PHG Foundation41 outlines fundamental challenges to using whole-genome analysis in a clinical diagnostic setting, including ensuring that the raw genome data are of sufficient quality and completeness and that the validity and accuracy of clinical interpretation is not undermined by an unreliable evidence base. The report stresses that using whole-genome data in clinical diagnostic services within the NHS without first addressing the fundamental quality-related issues poses potentially unacceptable risks to patient safety and quality of care. These risks include:

• incorrect diagnosis (false positives or false negatives) leading to inappropriate patient care and decision-making and threatening patient safety

• failure to provide a conclusive diagnosis for the patient and a continuation of their diagnostic odyssey

• inappropriate use of NHS resources.

As such, the authors recommend four steps to overcome these challenges: determining minimum standards for sequence generation; establishing best practices in bioinformatics; improving the quality of the evidence base; and building infrastructure. However, it is noted that implementation of these steps will be time and resource intensive and will require a methodical approach. 41

Further publications also highlight the need for a clear, evidence-based policy on how to conduct genome testing. For example, a paper on the policy challenges of clinical genome sequencing33 argued that a clear, evidence-based policy on how to conduct genome testing is both essential and urgent. It focused specifically on problems of data overload, incidental findings, widening public understanding and focusing on diagnosis. 33 Although this paper states that genomic sequencing is now sufficiently cost-effective to be offered clinically, interpretation of individual genomic variation remains challenging and the importance of incidental findings is unclear. Thus, it may be premature to offer opportunistic genome screening without knowledge of population penetrance. It is recommended that clinical genome sequencing efforts should initially focus on delivering diagnoses for patients, with efforts also focusing on developing a shared database of sequencing results linked to phenotypes to facilitate research. The authors also state that, in patients whose clinical presentation is indicative of a primarily genetic aetiology, a combination of genome-wide sequencing and clinically targeted analysis can maximise interpretable pertinent findings and minimise non-pertinent incidental findings. Furthermore, they recommend that guidance and educational material be developed for those involved in NHS genomic sequencing, particularly focusing on the purposes of and limitations associated with testing, the difficulties of interpretation and the need to share the data acquired through such testing.

Giving informed consent

Another issue associated with introducing NGS technologies into clinical practice relates to the ethical challenges and practical dilemmas associated with informed consent, particularly how incidental (or unsolicited) findings will be handled. Because NGS technologies are highly likely to generate incidental findings, it is suggested that guidance is needed on:

• the information that should be discussed and consented to preceding the test

• which results to disclose to the patient (which may have consequences for relatives)

• the extent to which patients should decide on what should be reported back to them. 42

Rigter et al. 42 also propose a set of questions to help gain insight into the degree of feedback that patients (or, in the case of children, their parents) should and could be able to decide on when the results from using NGS technologies in diagnostic settings are reported back to them. These are presented in Box 1.

In particular, the need for an optimal informed consent procedure for exome sequencing in diagnostics has been highlighted. 43 The 2013 policy paper of the European Society of Human Genetics on WGS concluded that, to develop best practice, information sharing structures must be set up and guidance must be put in place regarding clinical utility, procedures for obtaining informed consent and how best to handle unsolicited findings/variants. 44 This paper recommended that protocols be adjusted as knowledge and patient and public responses evolve, with a view to any current recommendations being re-evaluated in a few years’ time.

In the USA, the ACMG has published guidance regarding informed consent for genome/exome sequencing. 45 Given that incidental findings are managed differently in the USA from in Europe, the ACMG document is not directly appropriate for the UK context. However, it – along with similar documents, statements and reports published in Europe – reflects that much discussion and debate about patient rights and professional duties has taken place. Despite all of this effort and expertise, no firm conclusions have been reached, and such activities will continue at local, national and international levels. It is important that such debate and discussion is undertaken in the UK on an ongoing basis to ensure that equitable, consistent and evidence-based approaches are adopted for the provision of informed consent, thus safeguarding all stakeholders in this process.

Reporting test results and managing incidental findings

Research literature differentiates between pertinent findings (those that have been purposively generated or sought) and incidental findings (those that emerge during the course of an investigation and that may have potential clinical and/or health implications but which were not specifically sought out as part of a test). 46 When considering the disclosure of findings generated by WGS undertaken in a research context, five graded possible approaches have been identified:46

• non-disclosure of potential findings that might have an impact on the health of an individual

• disclosing only clinically significant findings that are severely and moderately life threatening and clinically actionable

• disclosing all clinically significant findings regardless of their severity and actionability

• disclosing all variants (regardless of severity, clinical significance and actionability)

• research participant chooses a disclosure policy.

It is considered part of a physician’s duty of care to his or her patients to assess the potential benefits and harms associated with the disclosure of additional findings, informed by knowledge of the patient and the clinical context. A PHG Foundation discussion paper46 identifies several aspects that the physician might consider when deciding what to disclose:

• the seriousness of the presenting problem and the nature of the other findings

• whether the finding represents a known clinical entity or risk factor or requires further investigation (e.g. variants of uncertain clinical significance)

• whether the finding has been validated to an acceptable standard

• the availability of any treatment/prophylaxis and its likely success

• whether the finding is a risk factor for disease or represents a disease process

• the age of the patient and co-existing morbidities and conditions

• prior knowledge of the patient’s wishes

• custom, practice and precedent: how other professionals have handled the same finding in similar circumstances and the extent to which there is consensus about this approach.

In a clinical context, it is suggested that it is the ‘professional obligations of beneficence and non-maleficence’ and the responsibilities of the physician that will drive the approach to the disclosure of additional findings, rather than the individual patient’s autonomous wishes. 46 Adopting this approach ideally requires physicians to have easy access to information about custom, practice and precedent. This, in turn, requires a mechanism or framework for collecting and presenting this information.

The UK10K WGS project, which took place between 2010 and 2013,47 identified the following four criteria that had to be met before a pertinent finding or an incidental finding could be passed on to the participant:

• Respect the research participant’s right to know or, alternatively, their right not to know about clinically significant findings. Explicit consent must have always been given by the participant for incidental findings and/or pertinent findings to be returned.

• Clinically significant findings should be returned only when they are of significant clinical importance.

• Clinically significant findings should be clinically and analytically validated by a genetically accredited laboratory before being returned to participants by their treating clinician.

• Feedback to individual participants should be conducted by a trained professional able to provide genetic counselling.

It may be appropriate to develop and adopt a similar ethical framework for managing clinically significant findings arising from using NGS technologies in clinical settings, although this would require strong leadership at a national level. If this is not forthcoming, there may be scope for developments at a regional level.

The 2013 paper48 on WGS in health care by the European Society of Human Genetics focused on the clinical diagnostics setting. Two of its recommendations are of particular relevance to handling incidental findings:

• when, in the clinical setting, either targeted sequencing or analysis of genome data is possible, it is preferable to use a targeted approach first to avoid unsolicited findings or findings that cannot be interpreted

• in case of testing minors, guidelines need to be established as to what unsolicited information should be disclosed to balance the autonomy and interests of the child and the parental rights and needs (not) to receive information that may be in the interest of their (future) family.

Finally, to show that there are no universally agreed approaches to addressing these ethical dilemmas, the work of the ACMG should be noted. The ACMG published its recommendations for reporting incidental findings in clinical exome and genome sequencing in 2013. 49 These recommendations require laboratories, regardless of the indication for which the sequencing was ordered, to explicitly seek and report on a minimum list of variants; patients are not allowed to decide whether they wish to receive such information. Instead, the ACMG recommends that patients who decline to receive information about these incidental variants should not be allowed to have their genome or exomes sequenced. Total responsibility is placed on the referring clinician by the laboratory geneticist. These recommendations are considered by some to be contrary to international ethical standards50 for a number of reasons, for example the issue of coercion has been raised with respect to not giving patients a choice of other options and denying them a test if they decline to receive information about secondary variants. It has been suggested that medico-legal considerations have played some part in their development. 50

Sampling and participants

A sample of stakeholders (n = 33) was recruited. In a scoping exercise such as this, identification of appropriate individuals to interview is necessarily pragmatic. In effect, a ‘snowballing’ approach was employed. Three key categories of stakeholders were identified initially:

• doctors working with children with LDs in a generic capacity in primary and secondary care settings [e.g. paediatricians and general practitioners (GPs)]

• genetics specialists working in laboratories and/or NHS clinical settings (e.g. clinical scientists, bioinformaticians, clinical geneticists, genetic counsellors)

• representatives from charities (or similar organisations) working with families of children with LDs.

A number of key individuals were identified by members of the RAG, through their recent publications or from web-based information about their work. Interviewees were drawn mainly from NHS settings, although several also had academic links with a university. Some also had a wider ‘national’ perspective from their activities with organisations such as the British Academy of Childhood Disability (BACD) or from their work on clinical trials. These individuals were also asked to suggest colleagues with relevant expertise or experience who might also be willing to be interviewed. This approach resulted in contact with a number of people from organisations with a national focus, such as the Department of Health, the Nuffield Council on Bioethics, Genetic Alliance UK and Nowgen (an organisation specialising in public engagement, education and professional training in biomedicine). An internationally renowned professor from Holland was able to provide a European perspective. A list of interviewees is presented in Appendix 4.

Interviews and data analysis

Semistructured interviews were conducted with all 33 participants. All but one of the interviews took place on a one-to-one basis by telephone; the remaining interview (involving two interviewees and two researchers) was undertaken face-to-face at a professional conference.

Each interviewee was sent a list of possible discussion topics before interview (see Appendix 4). Each interview commenced with the participant being asked about his or her understanding, knowledge and experience of NGS. The remainder of the interview was then tailored to reflect the knowledge and specialist area of each interviewee. Each interview concluded with a brief discussion of the participant’s views of the main strengths and weaknesses of traditional approaches and NGS, including their implications for patients. Interviews lasted between 30 and 60 minutes.

Interviews were audio recorded and transcribed verbatim. All interviewees were offered a copy of the transcription so that they could, if they wished, add any additional points that may have occurred to them after the interview had ended.

Interpreting the results

The views of the stakeholders are summarised in the following section; various quotes from the interviews are reported in Appendices 5–7 and provide an evidence base for the material in this section and the subsequent discussion in Chapter 5. Stakeholder views should be considered in conjunction with the material from published reports, papers and websites discussed in Chapter 2; the material was selected and presented to provide contextual and factual information about specific aspects and developments mentioned by interviewees. When interpreting the material from the stakeholder interviews it is important to note the influence of the varied backgrounds and experiences of the interviewees on their responses and perspectives. Clinicians, such as paediatricians and GPs with first-hand experience at a local level of caring for children with LDs and their families, generally tended to take a pragmatic perspective and consider what they believed would be needed to assist their clinical practice at a local and/or regional level. However, those with an overarching focus, such as individuals representing government bodies and charities, generally had a wider view of what would be needed/desirable from a national standpoint. Those working primarily in NHS genetics laboratories provided another dimension. Furthermore, some interviewees focused entirely on the potential use of NGS technologies for diagnosing children with LDs, whereas others considered the broader role of NGS technologies for diagnosing other conditions. Some had direct experience of using NGS technologies in research settings. A few, especially those working directly with patients in NHS settings, had little knowledge of NGS technologies but were familiar with some of the recent challenges associated with introducing aCGH into the NHS for diagnosing children with LDs.

Data-related issues

NHS staffing and training issues

Almost all of the interviewees highlighted the need for additional staff, especially in the short term. Although some of these personnel, especially those who will undertake diagnostic work in laboratory settings, may be recruited from staff already working on research projects, the need for increased numbers of trained staff working in clinical settings was frequently stressed, alongside the need for additional government-funded training courses, especially for bioinformaticians and clinical scientists. Several interviewees also referred to the need for wider training for clinical staff working with children with LDs (i.e. the ‘patient-facing’ professional, including nurses and GPs), to improve their knowledge and awareness of genetic issues.

Other aspects

Three other broad areas for changes in service provision emerged from the stakeholder interviews, namely the need for public education, the importance of family support and changes to service commissioning.

Giving informed consent

Participants highlighted the ethical challenges and practical dilemmas associated with informed consent, particularly with respect to the handling of incidental findings. Many of these issues were also identified and discussed in the literature review (see Chapter 2). In particular, several interviewees highlighted the problems of getting patients to consent to something unknown that may not even relate to their presenting condition.

Reporting test results: managing incidental findings

Participants identified that one of the most controversial aspects of WGS is the ethical issues that may arise from its use (or indeed from the use of other NGS technologies) in clinical and research settings. The research literature base pertaining to reporting test results is discussed in Chapter 2. Almost all of the interviewees mentioned this as an area where further work is needed. The preference was for the work to be undertaken at a national level but also to draw on the experiences of other countries.

It was explained that issues associated with the amount of information from testing to give patients are not new but are exacerbated by the extremely large amounts of data generated by NGS technologies. Different countries are taking different approaches to managing these issues, for example because of medicolegal considerations the approach in the USA is completely different from that being adopted in most European countries. The Americans state that incidental findings must be disclosed whereas the European approach is to undertake only targeted testing. It was suggested that these two views ‘represent opposite ends of the spectrum with Britain expected to be “somewhere in the middle” ’. Differences may also arise between diagnostic testing for research and diagnostic testing for clinical care. In reality, there are logical reasons and justifications for the different approaches being suggested in different countries.

Ethical and other frameworks

A variety of issues pertaining to ethical and other frameworks were raised during the stakeholder interviews. These are summarised in Box 2.

One interviewee raised the need for further research on false-negative rates and stated that ‘the implications of false negatives and false positives are very different’. This interviewee stressed the importance of identifying the false-positive rate over the next few years.

In addition, the importance of a regulatory framework was raised by some of the interviewees, as was the importance of public education and public debate. However, it was also pointed out by one interviewee that ‘you could never eliminate any possibilities of misuse unless you eliminate any possibilities of use’.

Pace, and extent, of change

It was considered that Holland (in particular) and Britain are seen as world leaders in the adoption of genetic testing for clinical work. Although the technologies are available in the USA and are being used for research purposes, it was suggested by some interviewees that their relatively slow adoption there for clinical work is because of the reluctance of insurers to fund them.

Interviewees had diverse views about timescales and pace of change for introducing NGS technologies into the NHS for diagnostic work. They recognised that the technologies already exist but estimated different timescales and end points. Most considered that the focus would be on targeted gene panels for the next 3–5 years but would shift to exome sequencing thereafter. There were fewer consensuses about any subsequent wholesale move from WES to WGS. Some interviewees based their views on their personal experience of working with children with LDs; others had a wider perspective on the clinical use of NGS technologies in the NHS. The variations in their views are reflected in the quotes included in Appendix 7.

It was generally considered that there would not be a national strategy for the widespread introduction of NGS technologies into NHS clinical work, resulting in developments occurring in an ad hoc, unco-ordinated way. With regard to introducing changes specifically for the clinical diagnosis of children with LDs, it should be noted that the technology for microarrays was available for several years before the NHS adopted it routinely as the first-line test instead of karyotyping. The same may also apply to the move to NGS technologies, especially without a national strategy. One interviewee neatly summarised one of the challenges of working in genetics thus:

Genetics . . . you think you understand it, you think you have got there, you think you have got your head round everything and you totally understand whether something is clear or not, and then things move on and you realise that actually things are even more complicated than you thought . . . . And even things that you thought were certain maybe are not quite as certain as you believed them to be.

Genetic specialist, NHS

Cost-effectiveness of comparator genetic testing technologies

The material in Appendix 3 shows that various studies have considered the effectiveness of microarray analysis (compared with other forms of testing such as karyotyping). However, only a few have considered the cost-effectiveness of such approaches. Information presented in this section highlights the fact that the studies that do exist were generally undertaken several years ago and not all took place in Britain.

A paper by Wordsworth et al. 25 published in 2007 considered the cost-effectiveness of microarray technology in the NHS for diagnosing ILD. The authors considered that, at that time, one obstacle to the adoption of aCGH in the NHS was concern over the proportion of confirmed genome imbalances for which the significance of the positive result was unknown (e.g. very small de novo imbalances and some inherited imbalances). However, they thought that the most significant obstacle was the perception that the technology was prohibitively expensive for use in most NHS cytogenetics laboratories. They therefore investigated the cost-effectiveness of aCGH compared with standard cytogenetic analysis for detecting genomic imbalances that diagnose ILD in the NHS.

They found that the average cost of aCGH was £442 per (single) patient sample and the average cost of karyotyping was £117 per sample (with the majority of the cost difference being accounted for by the array cost). Therefore, from a single-test perspective, aCGH was indeed more expensive than karyotyping.

However, the authors took the view that, in reality, the situation was more complex because information regarding subsequent tests for genomic imbalance must also be considered before estimates of true cost-effectiveness can emerge. They calculated that the overall cost per diagnosis of the karyotyping route, including a single multi-telomere FISH assay (£4957), was more expensive than the overall cost per diagnosis of the aCGH route (£3118), which yielded 10% more diagnoses. However, if the less conservative yield of 15% more diagnoses is correct, then the aCGH cost reduces to £2440 per diagnosis. This figure is more comparable to the cost of karyotyping plus the alternative multi-telomere assay, MLPA (£2129 per diagnosis). However, the authors state that 92% of cases tested by karyotyping and a multi-telomere assay will require further tests for a diagnosis. By contrast, they suggest that the aCGH route is unlikely to require further genome-wide tests for genome imbalance.

Wordsworth et al. 25 considered that their study had shown that it was viable for the NHS to use genome-wide aCGH for the diagnosis of ILD and that, in some cases, it might be appropriate to replace karyotyping with aCGH as the first-line test for genomic imbalance in ILD. They suggested that samples shown to be normal following aCGH could be karyotyped to identify truly balanced rearrangements or further characterise genome imbalances. Alternatively, aCGH could be useful for clarifying previous equivocal karyotyping results. They concluded that testing for genomic imbalances in ILD using microarray technology is likely to be cost-effective as an earlier diagnosis saves the cost of additional tests and a negative result minimises the choice of follow-up tests.

In the same year, in the UK, Newman et al. 10 published an exploratory cost–consequences analysis of using aCGH for the diagnosis of DD. They defined some of the resource cost implications of introducing aCGH into clinical practice in terms of avoiding or reducing the use of other investigations required to obtain a diagnosis. The medical notes for 46 consecutive patients selected for aCGH analysis by six experienced dysmorphologists at the Regional Genetics Service in Manchester were abstracted for all clinical investigations relating to the determination of the cause of LDD.

Newman et al. 10 found that the number of investigations undertaken on each child varied. With aCGH estimated to cost £590 per case, they found that the additional cost would be £2399 per positive case if aCGH had been undertaken after negative standard initial tests for LDD investigation. If the cost of aCGH was reduced to £256 per case, aCGH would become cost neutral. They also found that all chromosomal anomalies detected by aCGH had a de Vries score of ≥ 5. If aCGH had been used only for individuals with a score of ≥ 5, the cost would be £1087 per positive case identified and the sensitivity would increase considerably.

Newman et al. 10 also stated that the economic implications of introducing aCGH should be considered in the context of other diagnostic tests for LDD. For example, only about 0.6% of males undergoing fragile X syndrome mutation will be positive for a full mutation. In their laboratory, > 700 samples were screened for fragile X syndrome each year. The use of this simple, inexpensive genetic test (£63 per case) equated to a cost of > £11,000 per positive case identified. Despite these cost implications, fragile X screening for LLD was considered standard clinical practice. This suggested to the study authors that performing aCGH on selected patients with LLD would be ‘an appropriate intervention in both clinical and economic terms’.

With regard to the cost-effectiveness of diagnostic testing, Miller et al. 52 considered that, except in certain cases, for example those involving a family history of multiple miscarriages, a karyotype was not cost-effective in a child with DD/ID, ASD or MCA and a negative array study. They also considered that, although CMA testing was not inexpensive, the cost was less than the cost of a G-banded karyotype plus a customised FISH test such as subtelomeric FISH and the yield was greater.

Trakadis and Shevell53 used Canadian data to consider the cost-effectiveness of microarray as a first genetic test in GDD. They evaluated the cost-effectiveness of aCGH compared with karyotyping by retrospectively analysing the cost of work-up in a cohort of 114 children representing a consecutive series of children diagnosed with GDD. Their costing approach also included considering the costs of using different types of laboratories. In a similar manner to Wordsworth et al. ,25 they focused on the cost per diagnosis detected rather than the cost per life-year or quality-adjusted life-year gained, which are standard measures often used in economic evaluation. This was because the diagnosis of GDD is highly unlikely to save lives and evaluating quality-adjusted life-years is intrinsically problematic in children, especially those with GDD. They concluded that ‘aCGH would be cost-effective as a first genetic test in the clinical evaluation of individuals with GDD’53 (p. 994).

Lynch24 stated that, although the cost of array testing was initially prohibitively expensive, the cost has reduced over time. She considered that the test was now affordable and also saved costs as in many cases it results in an earlier diagnosis, thus removing the need for second-line diagnostic investigations, including invasive investigations such as brain imaging and muscle biopsy. She also suggested that, as they become established over time, ‘microarrays should result in fewer hospital appointments and fewer additional (now unnecessary) diagnostic investigations’24 (p. 971). However, she also stated (p. 971) that these ‘hidden’ benefits are difficult to cost for several reasons:

• the ordering of first and second-line investigations by doctors differs not only between countries but also between hospitals (arrays will bring less financial benefit the more first-line investigations are ordered)

• some children with an array anomaly may require additional investigations and it is difficult to estimate how many investigations can be avoided

• it is not unusual for paediatricians to routinely order Fragile X testing at initial assessment; however, it is hoped that, in time, these requests will only be activated should the array be normal

• many CNVs will be detected and it will require parental analysis to confirm whether each identified CNV is benign, thus adding to overall costs

• it is difficult to estimate health-care savings as, even with a diagnosis, these children will still require long-term support from health-care providers

• the majority of children will test negative on array and will need further investigation.

Lynch24 also mentioned the benefits associated with a diagnosis. These are explored in more detail in Appendix 6 but of particular note is the long-lasting emotional relief for parents and the fact that having a diagnosis generally facilitates the receipt of social and educational support.

Regier et al. 5 explored whether using array genomic hybridisation (AGH) for diagnostic testing of the genetic causes of ID provides value for money. The term AGH is used in the generic sense to encompass the class of array technologies that detect submicroscopic CNVs. The authors used decision analysis modelling to evaluate the trade-off between costs, clinical effectiveness and benefits of an AGH testing strategy compared with a conventional cytogenetic testing strategy. Using Canadian cost data for 2007/8, they found that the baseline AGH testing strategy led to an average cost increase of $217 [95% confidence interval (CI) $172 to $261] per patient and an additional 8.2 diagnoses in every 100 patients tested (0.082; 95% CI 0.044 to 0.119). The mean incremental cost per additional diagnosis was $2646 (95% CI $1619 to $5296). Probabilistic sensitivity analysis demonstrated that there was a 95% probability that AGH would be cost-effective if decision-makers were willing to pay $4550 for an additional diagnosis. The authors concluded from their analysis that using AGH instead of conventional karyotyping followed by FISH for most ID patients provides good value for money. Their model also showed that obtaining a karyotype for all children with ID and then testing with AGH if the cytogenetic analysis did not provide a diagnosis was ‘not cost beneficial when compared to using AGH as first-line diagnostic testing’5 (p. 771).

Cost-effectiveness of next-generation sequencing technologies

The costs of undertaking diagnostic testing using WGS and WES are steadily falling and the diagnostic yields are anticipated to improve over time as the technologies become more established. There are therefore no (rigorous) published studies comparing the costs and effectiveness of these technologies with those of more traditional diagnostic techniques.

Some work on the cost-effectiveness of NGS technologies has been undertaken in Holland, often focusing on WES and/or targeted gene sequencing. Heger31 reports in Clinical Sequencing News on the early experience of a group (led by Joris Veltman at the Radboud University Nijmegen Medical Centre) investigating the cost-effectiveness of exome sequencing.

Towne et al. 54 used American data to measure the cost-effectiveness and efficiency of WES/WGS compared with targeted gene analysis in a paediatric population with an undiagnosed potentially genetic disease. This work is in its preliminary stages. Genetic diagnoses had been identified in 22% of 45 families when initial findings were presented at the 2013 meeting of the American Society of Human Genetics. The authors concluded that their preliminary data suggest that, compared with targeted sequencing, WES/WGS is an efficient and probably cost-effective way of diagnosing the genetic basis of disease. However, they highlighted that their findings are reflective of research testing rather than testing in a clinical laboratory. Furthermore, as the study is in its early stages and is being undertaken in an American context, it is not currently clear whether its findings will be transferable to other countries.

Diagnostic test costs

In some cases it will be possible to make a diagnosis (with or without subsequent confirmatory gene testing) simply by taking a full medical history and examining the child. In other cases the medical history and examination will inform which tests should be carried out. For children who present without a suggested observation-based diagnosis, the paediatrician is likely to order a number of first-line tests. If these do not result in a diagnosis, a number of more specialised second-line tests may be requested. If the second-line tests still do not yield a diagnosis then third-line testing may be carried out on a proportion of children. A ‘representative’ list of potential tests and their NHS costs (Dr Kay Metcalfe, Central Manchester University Hospitals NHS Foundation Trust, 2014, personal communication) is included in Table 6; available information on the cost of genetic tests is provided in Table 7.

Genetic counselling cost

Information on the cost of nurse counsellors undertaking genetic counselling is included in a study that explores ways of improving the referral process for familial breast cancer genetic counselling. 59 This study, which took place in two locations (one in Grampian, Scotland, and one in Wales), assumes that breast cancer genetic counselling consumes a similar level of resource to other forms of genetic counselling. Costs are divided into four categories, namely staff, consumables, rooms and equipment. The study reports that staff estimated that they allocated approximately 1 hour per patient during the clinic appointments and that an additional hour was required for preparation. Furthermore, there were various indirect costs. These were largely for meetings, such as weekly clinic meetings, during which all genetic patients were discussed. In addition, in both study centres, nurse counsellors were required to meet a consultant geneticist to discuss patient cases and determine the need for any further counselling and/or management recommendations (30 minutes in Grampian and 90 minutes in Wales). In the control arm in Grampian, three specialist registrars each met with a consultant geneticist for 20 minutes per month to discuss patient cases; in the control arm in Wales, a clinical assistant met with the consultant geneticist for approximately 1–2 hours each month. The staff, consumables, rooms and equipment costs are displayed in Table 8 at 2001 prices, with costs converted to 2013 prices using the Hospital and Community Health Services Index. 60

Establishing the value of using next-generation sequencing technologies to diagnose learning disabilities

Next-generation sequencing technologies are used not only for the diagnosis of LDs but also for the diagnosis of a plethora of other conditions, including cancers, cardiac problems and neurological conditions. It is therefore difficult to estimate how much of the value of NGS technologies should be attributable to their role in diagnosing LDs and how much should be attributable to other roles. Furthermore, as LD panels are still in their infancy and WES and WGS are currently being used only in research settings, it is difficult to determine the value of NGS technologies to the NHS.

Establishing the effectiveness of next-generation sequencing technologies

Finding good evidence regarding the performance of diagnostic tests and interpreting its value for practice is more challenging and less straightforward than for interventions. 61 Most diagnostic studies focus on a test’s ability to discriminate between people with and people without a particular condition. In the case of genetic testing for LDs the case is complicated because the test has the potential to identify not just one condition but a number of different conditions. Furthermore, having a highly accurate test does not necessarily improve a patient’s outcome, although it may avoid (or limit) the necessity of carrying out further investigative procedures.

For children who present without a diagnosis the paediatrician is likely to order a number of investigative tests. There is currently no information on the number of children who receive each available diagnostic test or the diagnostic yield of these tests. The issue is complicated by the fact that diagnostic yield is influenced by the position of the test in a patient’s diagnostic pathway, which in turn may be correlated with the time span over which a diagnosis has been sought.

Identifying an unambiguous diagnostic yield for a particular technology is far from straightforward, if not impossible. A key factor is that comparing findings from different studies is difficult because of the differences in technology used as well as the differences in the characteristics of the patient groups involved in the studies. The current situation is that, although there are numerous studies reporting diagnostic yield for karyotyping and aCGH for specific populations, it is difficult to generalise results to the population with LD with any degree of certainty. There is even less certainty with regard to the diagnostic yields from NGS technologies.

Three particular limitations of NGS technologies have been highlighted (Professor Frances Flinter, 2013, personal communication):

• NGS technologies generate poorer quality of reads than Sanger sequencing, producing an error rate of 1–1.5%.

• Errors in machine or panel manufacture lead to false-negative results. The false-negative rate for NGS technologies is unknown but it is thought to be high and to be the greatest problem for NGS technologies at the present time.

• The technology has yet to accurately identify CNVs (insertions/deletions/duplications) because of the short reads. Methods for detecting CNVs do not meet the standards required for a clinical diagnostic test, for example reads that fall in repetitive regions of the genome cannot be mapped unambiguously to the reference sequence.

Furthermore, although using NGS technologies to generate data is relatively straightforward, the analysis of that output to produce clinically useful information is still an evolving area of expertise. Until diagnostic yield, or another efficacy measure, is established for NGS technologies and their comparators, it will not be possible to establish effectiveness with any degree of certainty.

Drummond et al. 62 have published a paper that includes details of particular challenges faced when attempting to establish the cost-effectiveness of devices. The authors of this paper note that those undertaking economic evaluations are often quick to ‘genericise’ their recommendations unless there is specific evidence to differentiate products. Whether or not this is an issue in relation to using NGS technologies to diagnose LDs is unclear. WES and WGS appear to be a ‘step change’ from more established forms of genetic testing. However, sequencing machines made by different manufacturers do not generate identical output data and therefore, if a study comes to conclusions based on the output from one manufacturer’s machine, these findings may not hold with respect to practice using a machine made by a different manufacturer. In addition, as technology in this field is developing very rapidly, it may not be possible to generalise study findings from an earlier version of a machine to those from a later, updated version.

Impact of a diagnosis on the NHS

From a NHS perspective, a diagnosis can be cost-effective only if it leads to improved treatment, which, in turn, leads to improved outcomes for patients. The extent to which care and treatment may be modified as a result of a diagnosis is very variable and is likely to depend on the specific diagnosis. In some cases there will be no change in treatment, with clinicians simply continuing to treat the symptoms. In other cases the diagnosis may result in ongoing monitoring for, say, heart disease and/or changes in the availability of social care provision and educational support. If there is no change in treatment as a result of a diagnosis then there may be no cost savings to the NHS other than those associated with ongoing attempts to obtain a diagnosis.

In addition, it is important to consider that a diagnosis sometimes leads to further tests and that some of these tests need to be repeated throughout life. So, for example, a diagnosis of Down syndrome would be accompanied by an appointment for an echocardiogram, if the child had not already had one, and would lead to regular thyroid function tests (minimum of 2 yearly throughout life), regular hearing and vision checks and a low threshold for investigating haematological malignancy or coeliac disease. However, the extent to which an individual might have some, or all, of these tests despite not having a diagnosis is unclear.

The extent to which further confirmatory genetic testing may be carried out following a diagnosis depends on the condition that has been diagnosed. Similarly, the extent to which the child’s diagnosis leads to genetic testing (and counselling) of family members is condition specific. In the case of family members the issue is complicated by the fact that family members’ wishes will also influence the extent of any testing (and counselling).

All of these tests (and any counselling) have supporting activities (e.g. outpatient appointments) and the cost of these needs to be taken into account in any economic evaluation. In addition, if an economic evaluation takes a perspective wider than that of the NHS, then the average cost to families of attending such appointments (as well as any costs associated with not having a diagnosis) needs to be quantified.

Impact of a diagnosis on the family

Although the value of a diagnosis was not explicitly included in the initial brief for this scoping study, several of the interviewees expressed their views on the importance of a diagnosis to the families, clinicians and other professionals. These views, which are not necessarily comprehensive, are provided in Appendix 6. Some views from the published literature are included in Tables 3 and 4. There is currently no published quantitative information on the benefit of a diagnosis to the family of a child with LD. In addition, there is a lack of information on the cost consequences of a diagnosis (or lack of one) for subsequent family planning decisions.

Wider economic implications of using next-generation sequencing technologies to diagnose learning disabilities

The introduction of NGS technologies has wide economic implications in terms of both technological infrastructure and training and education.

With regard to technological infrastructure, feedback from the interviewees suggests that the biggest issue relates to storing the huge amount of data that the use of NGS technologies can generate. This has implications in terms of needs in relation to both computer hardware and computer software. There are also associated considerations around data confidentiality and the length of time that data should be stored. It has been suggested that the costs associated with processing, transfer and storage issues may be more than the costs associated with producing the data (Professor Frances Flinter, 2013, personal communication). These are all costs to the NHS, which are difficult to quantify. Moreover, these costs are likely to change substantially over time. Although these costs are very uncertain, ignoring them underestimates the cost of introducing NGS technologies into the LD diagnostic pathway.

Some storage costs could be avoided if a policy of not storing data (but rather resequencing on occasions when it was deemed appropriate to revisit a child’s genetic code) was introduced. However, there is still currently a need to store considerable amounts of data for research purposes, that is, so that it can be used to help researchers distinguish between genetic abnormalities and normal variation.

In terms of training and education, information from the interviewees, as discussed in Chapter 3, suggests that there would be a need for more bioinformaticians and more genetic counsellors. In addition, interviewees considered that a wide range of health-care professionals would need to know more about genetic testing.

Estimating production and non-production costs associated with next-generation sequencing technologies

Production costs

The production costs of NGS technologies appear to be changing on an almost daily basis. For many years the National Human Genome Research Institute (NHGRI) in the USA has tracked the costs associated with DNA sequencing performed at the sequencing centres that they fund. They present data since 2001 for two metrics: the cost per megabase (a million bases) of DNA sequence and the cost per genome. These data are shown in Figures 1 and 2 respectively. 63 Both graphs use a logarithmic scale on the y-axis. It is noteworthy that the sudden outpacing of Moore’s law (the normal expectation of the rate of change in the cost of technology over time) beginning in January 2008 corresponds to the time when the sequencing centres transitioned from Sanger-based to next-generation DNA sequencing technologies.

FIGURE 1.

Cost per raw megabase of DNA sequence. Note: Moore’s law is the observation that, over the history of computing hardware, the number of transistors or integrated circuits doubles approximately every 2 years. Technology improvements that follow the same trajectory as Moore’s law are widely regarded to be doing exceedingly well. Source: Wetterstrand. 63 Reproduced with permission.

FIGURE 2.

Cost per genome of DNA sequence. Note: Moore’s Law is the observation that, over the history of computing hardware, the number of transistors or integrated circuits doubles approximately every two years. Technology improvements that follow the same trajectory as Moore’s Law are widely regarded to be doing exceedingly well. Source: Wetterstrand. 63 Reproduced with permission.

Furthermore, the PHG Foundation report Next Steps in the Sequence28 points out that:

• new technologies have an associated ‘learning curve’ whereby the efficacy improves over time

• laboratory protocols and the knowledge created from the use of the technology are constantly evolving

• there are new advances in knowledge capital being made that have an inherent and tangible value which needs to be taken into consideration in any economic assessment, especially if this knowledge is not publicly shared and must be paid for through normal market-based exchange.

The authors of this report conclude that, over the next few years, as the technology continues to evolve, it is unlikely that there will be a period when the technology is sufficiently stable to allow meaningful comparisons over a given time frame. 28 An added issue is that the stabilisation of current NGS technology prices may (at least in part) be because they are being replaced by ‘next NGS’.

Other considerations

As well as NHS and personal social services, a person with LD may need help from other services, including education, housing, respite care and the benefits system. Furthermore, informal carers may carry significant burdens not only in terms of time input but also in terms of private expenditure. In addition, there may be further costs associated with loss of productivity as a result of unemployment or absence from work by carers. Moreover, knowledge of genetic risk carries potential harm and benefit to family members. 64 All of these costs are very hard to measure, primarily because they are difficult to generalise.

When designing an economic evaluation careful consideration needs to be given to the perspective of the evaluation. There is a trade-off between including all relevant costs and benefits (i.e. a societal perspective) and having to make large simplifications and/or very broad assumptions because of uncertainty and taking a very narrow view that focuses only on known costs and benefits.

A similar trade-off needs to be made when considering the time frame for the economic evaluation. One extreme would be to focus on the short period over which costs and benefits can be measured. The other extreme would be to adopt a lifetime horizon (in line with the National Institute for health and Care Excellence reference case65) and include bold assumptions and generalisations. A compromise between these two positions may be the best option.

The cost-effectiveness of next-generation sequencing compared with traditional genetic testing for the diagnosis of learning disabilities in children

Data capture by the laboratories

There did not appear to be a strong consensus among participants about the most appropriate machine/platform to use to test for specific conditions. Indeed, different platforms may be required for the different conditions associated with LD or possibly for research and diagnostic use:

The technology is out there, but if I’m honest, it’s still a bit embryonic. There are a lot of different technologies, so we haven’t yet arrived at a single platform, so there are a lot of other technology validation and technology assessments going on to work out whether there is a single platform that is best suited to our needs. It will change a lot over the next few years; there is no doubt about it. We are currently in what we call ‘second generation sequencing’ platforms. I don’t think it will end in second generation platforms; we will have third generation platforms.

Genetic specialist, NHS

Although it was generally considered that the data capture process was relatively straightforward, interviewees stressed the importance of it being carried out properly:

There is an ever-increasing discrepancy between the research domain and the NHS. . . . There are a very large number of genes that have been identified that cause LD and we should be [testing for them] now. What we need is money; we need an infrastructure of machines to do it. We need information technology to store data and we need human beings to analyse the data. But we also need to do it in a cost-efficient way, and be mindful of the long-term sustainability; the technology is moving along at such a pace that it would not be efficient to have multiple places doing it. One would imagine that we would need a few core, well-funded institutions that would have the infrastructure to replace and modernise as the new technology emerges. It’s perfectly possible; it just needs money and commitment to do it.

Genetic specialist, NHS and university, with a national perspective

With regard to laboratory provision, views differed about the extent to which this would be achieved by the NHS and/or the private sector. One interviewee expressed strong concerns that ‘the NHS has a notoriously poor history of private sector commissioning’ and favoured using NHS laboratory provision if possible. Another interviewee considered that:

Private labs are certainly on the horizon. Currently there are a number of private concerns that are manoeuvring into position to look at this. So at the moment private labs aren’t doing NGS for UK patients, but I can see it coming on the agenda very quickly.

Clinical geneticist, NHS, with a national perspective

Data interpretation

A biometrician’s work was described thus:

The day-to-day profile is about pipelining, and building robust pipelines to put the data through the systems in the diagnostic labs, as well as for the clinical research work that is being done. The targeted panel approach (which I believe LD is more connected with) is where you select certain genes that are known to be associated with LD. You [i.e. a bioinformatician] would use NGS technologies to try and ascertain all the variants across these known genes, and the clinical scientist would then try to come up with some sort of an interpretation based on any of these variants that have been known previously to be associated with a mutation causing LD in this particular instance.

Genetic specialist, NHS

Another interviewee working in a NHS genetics laboratory said:

You need the infrastructure and the machines to do the sequencing. You also need computing infrastructure and bioinformatics support, that’s very important. That’s one of the most important unticked boxes – getting people to do the bioinformatics and having the computer equipment to do it.

Genetic specialist, university and NHS

An interviewee working in a NHS regional genetics service considered that:

In theory, the generation of the test data is going to be relatively straightforward. The complex thing is the data storage and interpretation. . . . And most of the people who are in laboratories, be it biochemical, genetic or whatever, they have not been trained as bioinformaticians, they have been trained as wet lab scientists. It’s completely different training that you need.

Genetic specialist, university and NHS

Another interviewee stressed the importance for data analysis of knowing about the clinical presentation (or the phenotype) of the patient:

It’s the clinical input that people overlook, but yet without that clinical input the bioinformaticians are not going to be able to produce any data that is much use to anybody. . . . Data need to be analysed really carefully, and you can’t analyse it properly without knowing about the clinical presentation of the patient. . . . So the interpretation of all those data requires not only bioinformaticians, but molecular biologists and clinicians who know what those particular diseases look like and can help make a call as to whether there are variants in a particular gene that are likely to be significant or not and can help steer people to look at the right genes in the first place.

Genetic specialist, university and NHS, with a national perspective

Data storage

One interviewee discussed the importance of adequate computing capacity for both data analysis and storage:

Data storage and sharing of information is a huge problem for the NHS, and most genetics labs will not have the infrastructure, computing wise, to deal with this particular problem. And even if they have got it now, it’s certainly going to be a problem in the near future, as we move from panel-based work, limited to dozens or maybe even hundreds of genes, up to whole exomes and whole genomes.

Genetic specialist, NHS

Data storage was considered to be the bigger of these two problems:

Because each trust acts as an individual entity, we have to be responsible for our own data, we have to store our own data. There may come, eventually, solutions around secure cloud-based data sharing, but clearly that is for the future, it’s not here yet. So at the moment most of us are looking for local solutions, storing huge amounts of data on local systems and having to build our own.

Genetic specialist, NHS, with a national perspective

Data sharing

The need for better data storage facilities was also stressed. One interviewee with an academic and research perspective was ‘less concerned about whether the labs are NHS or private sector’ than about ‘access to the data, ownership of the data, and analysis of the data across all health service-recruited samples’. In their view:

What would be catastrophic is if the process of analysis for LD was fragmented into different people doing it and there was no central access to the data. . . . So what is most important over anything else is that the NHS data should be held in a central place that is freely available to other health service users.

Genetic specialist, university and NHS, with a national perspective

Several others also pointed out that it is important that those interpreting the data can access appropriate comparative data to support their work:

Testing and interpretation can be done in the same place or in different places – but research has made it clear that data interpretation predicates on access to large data sets, so it isn’t sensible if everyone is doing this on their small, discrete, little families. Powerful data interpretation requires having access to large data sets. A certain amount of data analysis can be automated, suggesting a centralised system as the most efficient.

Genetic specialist, university and NHS, with a national perspective

Staff training: testing and interpretation

The Manchester-based course for bioinformaticians was seen as an exciting development in the use of NGS technologies across the NHS, not just for children with LDs:

Without doubt, NGS is going to be used across the entire pathology discipline, and it’s already being used in different ways in microbiology, virology and ‘across the piste’. I think that clinical bioinformatics will have to be part of all pathology disciplines, so I think we are going to really need people who can give us the information in a format that can be interpreted. I’m not quite sure where the interface between the data and interpretation is going to lie – i.e. between the bioinformatician, clinical scientist and clinician. But obviously we are going to need loads more bioinformaticians.

Interviewee with a national perspective

Staff training: clinical staff

Many interviewees referred to the need for widespread ongoing training in genetics amongst NHS staff caring for children undergoing diagnostic testing for LDs and their families:

We need to look at the capacity of the health service to use the information from NGS appropriately. That means investing in infrastructure that would make speedy and accurate diagnoses scientifically and technically. But it also means investing in education and continuing professional development so that doctors and nurses, nutritionists, etc. who will have sequence information coming into their lives have the skills and support to be able to use the information appropriately.

Interviewee with a national perspective

We need a better-educated workforce – about genomics in general (including its benefits and limitations) – across all health and social care services.

Genetic specialist, NHS, with a national perspective

The wider workforce – secondary specialists, the whole primary care sector, pharmacy, nursing, dieticians, etc. – need to start to understand what genomics is – and how it differs from genetics. This is a real challenge for Health Education England – but programmes are in place and/or being developed.

Interviewee with a national perspective

Some interviewees stressed instead the importance of good team or partnership working and/or of needing to have knowledge only of specific aspects. One participant considered that:

At the moment, the way I work in partnership with my clinical genetics colleagues is that they would deal with the interpretation and the explanation of the ‘spelling mistakes’ in the gene coding. I would very much deal with the day-to-day management of the child’s secondary disabilities and troubleshooting for any additional medical complications that might be predicted because we know what the diagnosis is. So I think we are quite well off here in having that arrangement, but don’t think all localities have it.

Paediatrician/GP with a national perspective

This interviewee also thought that it was important that there are neurodisability-trained paediatricians in every locality ‘so that all the management of the secondary disabilities and all of the surveillance that is required can actually get done’.

A medical geneticist interviewee said that:

A laboratory clinician will do the interpretation, but before I tell someone I need to feel that I agree with the results. It depends how much you trust your clinical scientist, therefore there has to be a strong relationship between the doctor and the clinical scientist.

Genetic specialist, university and NHS, with a national perspective

When asked about the importance of education about genetics for clinicians, a general paediatrician considered that:

[Although] it would be great, because you could have someone giving you hundreds of talks on it, and explaining the nitty-gritty and it would be interesting to know, but in terms of practicality as a front-line clinician, you just need to know that it’s a genetic test, and that there is someone looking at it and interpreting it and saying ‘this could be significant’ or ‘no, it isn’t’.

Paediatrician/GP

One interviewee thought that teaching about genomics should be included in medical education and thereafter provided through CPD. In addition, this interviewee thought that a significant proportion of the current workforce need training (or at least considerable awareness raising) in this area. There would appear to be considerable scope for appropriate organisations to develop a range of different training media for this, including web-based opportunities. As one interviewee stated:

Classroom things are always good; and I think online is fine. It is very effective when the problem arises to have access to information. I mean, you can educate people around the area, but until they come across the need for it, I suspect they won’t really engage fully.

Paediatrician/GP

Another interviewee mentioned the effect of NGS on the process of consenting patients and handling findings of uncertain significance, both of which have a significant educative element:

[With NGS] a lot of education will be needed. You will generate a lot of findings of uncertain significance. So we will need a bigger work force, and education for people doing the consenting of these patients. It won’t be just the domain of the clinical geneticists; the paediatricians will be doing a lot of consenting. They will need education, and they will need more time.

Genetic specialist, university and NHS

An interviewee with an international perspective considered that it was important to have more rare disease specialists, as increasing numbers of new rare diseases would be diagnosed with NGS technologies:

For these rare diseases, we need clinical geneticists. At the very least, we need rare disease specialists. Whichever way the technology goes, that’s where the investment should be. Actually, more than in the bioinformatics. In the long term, I think the bioinformatics will sort itself out.

Genetic specialist with an international perspective

Genetic counsellors

Mixed views were expressed about the need for more genetic counsellors, although most of the interviewees considered that many more of them would be needed with the introduction of NGS technologies:

There are two reasons why you need more counsellors; the first is because this test is much more widely applicable than previous tests. And that means that you will be doing more testing, and so more patients get tested. So you need more counsellors because more patients get tested. But you also need more counsellors because the process of counselling for these complex tests is more involved, so it actually takes longer to counsel for these more involved tests than it would for a very simple test. . . . So I do think that more counsellors, or more counselling time, are required.

And will the counsellors need training in a wider number of areas?

Absolutely, I think more so than the scientists. The counsellors will need training on how to inform patients of the implications of this testing. That certainly has been our experience of rolling out the current testing we are using, that counsellors find it quite hard to understand the implications, the full implications. This is partly because you don’t know what the implications are – you sometimes can’t predict them going in to the test.

Furthermore, as the use of clinical NGS becomes more widespread, current counsellors would need to undergo the relevant genetics training to enable them to work effectively with families experiencing these new technologies:

Finding the time and the CPD to get existing counsellors trained when they are already very busy is going to be an issue and a problem.

Genetic specialist, NHS

A genetics counsellor explained that the delivery of genetic counselling may change considerably:

If we can test for lots of different conditions once [i.e. with NGS], then hopefully families won’t have a diagnostic odyssey. And that may have implications for how genetic counselling works. . . . Rather than putting the focus and attention in the very early stage, genetic counsellors are more likely to pick up families after they have got their diagnosis, to help with the adjustment, to help with the understanding. Because probably the results you are generating are going to be more complex if we are testing for lots of different things at the same time.

Genetic specialist, NHS, with a national perspective

This in turn raises additional problems:

One of the concerns for patient groups, and perhaps for [genetic counsellors] as well, is that those types of models where we are providing more post-diagnostic support will be great for the families that get a diagnosis, but where it might fall down is that it won’t give the same level of support to families that don’t end up with a diagnosis from any of these tests. And we won’t know if that’s because they don’t have a genetic diagnosis or the technology just hasn’t got there yet. So if all the NGS tests that we are able to throw at families all come back normal, it does leave this group of individuals without a diagnosis and without support.

Genetic specialist, NHS, with a national perspective

An alternative view was provided by a paediatrician interviewee, who works closely with the local clinical geneticists. This interviewee considered that there was possibly a greater need for more nurse specialists than for more counsellors, especially in places where clinical networks are less well developed. When asked if more genetic counsellors were needed, the interviewee replied:

I don’t think so. Certainly in our patch we cope across the consultant clinical geneticists and myself and my team. . . . I think they bring their nurse specialists to certain conditions and it may be that more nurse specialist posts will be developed around particular families of conditions as more knowledge becomes available. And then it would become more highly specialised and then would have resource implications. So, at the moment, I know in our area that there is a clinical nurse specialist for neurofibromatosis and [one] for fragile X syndrome. . . . But you can see in the future that there may well need to be other nurse specialists with particular niche expertise that would need to give extended support, perhaps where networks are less well developed.

Paediatrician/GP with a national perspective

Another interviewee (a consultant community paediatrician) described how their service often uses specialist health visitors to support families:

Within my service, I have specialist health visitors who offer support to families, so that would be our normal system. But if it was a condition that we were very unfamiliar with, we would liaise with the genetic counsellors around that.

Paediatrician/GP

Much of the above discussion was summarised thus by a professorial interviewee:

Medics that can interpret clinical history and molecular results are going to be gold dust, because everyone is going to want these genome results and no-one is going to know quite how to interpret them. We either need an absolutely huge education campaign with lots more clinical geneticists, or we need many more clinical geneticists working with doctors. I would say that the new breed of counsellors is going to be able to take on quite a large amount of that work. It needs genetic counsellors who are highly trained in molecular genetics as well as the clinical skills. And it is a big range of needs, going from data interpretation to explaining to patients. . . . You also need local leaders who are going to push things forward, who are going to tackle the issues and difficulties you have with developing genetics services. In addition, there has got to be a close relationship between the scientist doing the high throughput sequencing and the people interpreting it. We are going to have to have science centre hubs close to the local clinicians who are going to interpret the result. I think the regional system is pretty good for these hubs.

Genetic specialist, university and NHS, with a national perspective

Need for public education

Concern was expressed by one interviewee that the private sector may become involved in promoting WGS to the public. Another interviewee, however, thought that, even with clever marketing, such private sector involvement was unlikely to take off, partly because of the problems of producing any meaningful data for the client (especially in the absence of knowledge of the person’s phenotype):

But without some clear purpose for WGS, i.e. for somebody without any clear clinical question, it is likely to be almost impossible to find useful information. I doubt that it’s going to be a problem – the fear of the ‘genetic iceberg’, as it used to be called, is probably a bit misplaced. Having said that, if WGS does become popular amongst the ‘worried well’ – ‘recreational’ genetics was how one colleague in the DH [Department of Health] described it – it is clear that any interpretation will have to be done before results are handed to a GP, who could not reasonably be expected to interpret it.

Interviewee with a national perspective

Importance of family support

Several respondents, especially those working directly with children undergoing diagnostic testing for LDs and their families, stressed the importance of organisations offering them support and information. One respondent, when discussing problems associated with finding out more about some conditions, said:

I think Unique’s rare syndrome support has done a wonderful job in pulling together the literature on these new, rare microdeletions and duplications [i.e. the ones with alphanumeric names rather than being named syndromes] and developing leaflets. They’re fantastic, and that does give people some information.

Genetic specialist, NHS, with a national perspective

A representative from Unique stated that:

The increasing diagnostic capabilities [of microarrays] are brilliant for us, but there is also an implication that at some point we are a bit nervous that we won’t be able to cope with the increased numbers. It’s difficult to get a really good handle on how many more families there will be and what the real impact is. And we are just sort of waiting to see what happens with DNA sequencing.

Interviewee with a national perspective

Furthermore, Unique gets information about diagnoses only from the families and asks them to get a copy of their genetics report from their doctor or geneticist if they have not been given one:

I think that what we are looking at in the near future is combinations of [traditional testing and NGS], but probably eventually it will just be NGS. What we feel a bit nervous about with NGS is the ability to interpret the results. And about our ability to interpret the results, because we don’t really know what families are going to be given, so we don’t know what we are going to be dealing with. So it’s a bit of unknown.

Interviewee with a national perspective

Changes in service commissioning

One interviewee hoped that the development of directly commissioned services at a national level would be a great improvement on recent arrangements:

I think you have to look at the structure the NHS is creating for directly commissioned services, those that are not appropriately dealt with through the local commissioning services. It’s all about developing a level playing field for those who are in need of specialist services. Historically, under regional specialist commissioning groups in previous arrangements, there was significant regional variation in terms of access to specialised services. NHS England is claiming that it will put systems in place that will ensure that patients who need specialist services are able to access them on the basis of need rather than geography.

Interviewee with a national perspective

Another interviewee, considering the bodies with overall responsibility for developing awareness in the NHS about the potential of NGS technologies, summarised several themes thus:

The Royal Colleges have a big role to play in education and making sure the understanding is there and making sure the professionals know what to expect and how to deal with that. NHS England has a big role to play in ensuring that the NHS is ready for that, and that includes service reconfiguration and looking at workforce redevelopment. And also, as we move into the new world of nationally commissioned specialised services, that CCGs [Clinical Commissioning Groups] and others have an understanding of the importance of NGS in the diagnostic pathway, and how that can be accessed and what it can deliver.

Interviewee with a national perspective