Genomics—an aid to diagnosis not a replacement

Non-coding DNA

The 100 000 Genomes Project was set up in 2012 to sequence 100 000 whole genomes from NHS patients with rare diseases and cancer to identity genes implicated in their conditions. Initially patients were drawn only from England, but the project is now UK-wide.

Every cancer patient recruited contributes two genomes for comparison (one from a healthy cell, one from a tumour cell), and the genomes of patients with rare diseases are, where possible, compared with those of their parents (trio genome comparison).

Unfortunately, genomics is rarely as simple as a single gene determining whether a person is susceptible to a particular disease. Genes tend to work together, and their activity is influenced by a variety of environmental and other factors.

Furthermore, whole genome sequencing has shown that the 20 000 genes coding for proteins represent less than 5% of the genetic information in the human genome. About half of the remaining DNA is unique sequence and the other half is repeated sequence, and all this information has an important role in influencing, regulating, and controlling the expression of the genes.

The 100 000 Genome Project has focused almost exclusively on interpreting the protein coding sequences, says Dominic McMullan, head of the germline programme at Birmingham Women’s and Children’s NHS Foundation Trust. “It is incredibly challenging to look at non-coding sequences, and we are going to need hundreds of thousands of trio genomes for different conditions before we really start to delve into what could be going on in non-coding sequences,” he told the meeting, so understanding its role is some way off.

Rare diseases and cancer

The 100 000 Genome Project has so far sequenced 50 000 genomes, the majority from patients with rare diseases and their relatives. Most of these patients have had genetic testing but it has not given a diagnosis. Whole genome sequencing typically shows that around 400 of the 20 000 coding variants in their genome are rare and functional. Applying a gene panel of variants seen previously in patients with similar conditions highlights any likely relevant variants (usually 10-20) and, finally, a comparison with the genotypes of unaffected parents allows rare inherited variants to be excluded to pinpoint any genes of likely diagnostic importance.

Caulfield says that in 23% of cases, the analyses highlight actionable variants—for example, indicating that a certain treatment will or will not be effective. He adds that evidence suggests “that revisiting your data annually could raise diagnostic yield by a further 20%” as more is learnt about the role of specific genetic sequences.

Whole genome sequencing has proved more challenging in cancer. A sample of tumour tissue is taken either when the tumour is excised or by biopsy but needs to contain sufficient cancer cells and be preserved appropriately. Traditionally, samples have been fixed in formalin and embedded in paraffin (FFPE) to preserve cellular architecture for pathology, but this damages the DNA, so protocols now require part of the tumour sample to be “fresh frozen.”

It is important to do this in all suspected cases, emphasises Nirupa Mirugaesu, consultant medical oncologist at St George’s University Hospitals NHS Foundation Trust and clinical lead for molecular oncology at Genomics England, so that the tissue can be sent for sequencing if cancer is confirmed. “It is just not practical or feasible to get consent from everybody upfront for genomic testing that may not be applicable to them,” she told the meeting.

So far around 132 “potentially actionable genes” have been identified in cancer. This means that they contain small coding variants for which there is a reported or potential therapeutic or prognostic association and that could make a patient eligible for a particular treatment on the NHS or to enter a clinical trial.

Reducing overtreatment

Sequencing a genome now costs less than £1000 (€1100; $1400), and this will continue to fall as the process is automated further, Caulfield says.

Sequencing used to take months, but fast track samples are now being returned in an average of 12 days, and the aim is to achieve a median turnaround of six weeks for all samples. The target is six weeks because that is how long a patient usually waits between surgery and starting adjuvant chemotherapy, and the genomic results might influence the choice of treatment.

“We know that we are massively overtreating our patients,” Mirugaesu says. “In the common tumour types such as breast or lung cancer sometimes you may treat 100 patients for five patients to have benefit. The hope is that in due course we will have better predictive and prognostic identifiers” so that patients will not need to be exposed to the adverse effects of chemotherapy if they are unlikely to benefit.

Routine NHS data from hospital records are also being aligned with the patient’s genomic data “to get greater value” from them, Caulfield says. For example, analysis may show that someone with a neuromuscular disorder who received intensive physiotherapy lived years longer than a patient with an identical disorder who received none.

The UK has already become the first country to implement whole genome sequencing technology for routine diagnosis, antibiotic resistance profiling, and surveillance for tuberculosis. Over 3000 multidrug resistant Mycobacterium tuberculosis isolates have been sequenced, and clinicians are now receiving a drug resistance profile within 24 hours of submission, far faster than conventional cultures can provide it, Caulfield says.

“Generation Genome”

Last year’s annual report from the chief medical officer for England, Sally Davies, Generation Genome,2 focused on how genomics will be rolled out in the NHS.

“Patients with cancer or a rare disease should have access to genomics based care, and health and care professionals should consider this as a standard part of their approach,” Davies said. Patients with a rare disease currently face consulting five doctors, receiving three misdiagnoses, and waiting four years before their condition is finally diagnosed, which is distressing for them and a waste of NHS resources.

Her report says genomics should be embedded in all elements of the NHS, from screening to treatment, and achieving this will require genomics training to be provided to the next generation of doctors and offered to existing ones.

“There are large numbers of people to educate and train, but there aren’t huge numbers of people with the requisite skills and knowledge to impart that to their mainstream colleagues,” says Katrina Tatton-Brown, consultant in clinical genetics and reader in clinical genetics and genomic education at St George’s University Hospitals NHS Foundation Trust, so online courses will be important.

She outlines three short courses for continuing professional development available on the FutureLearn platform: an introductory course, “The genomics era: the future of genetics in medicine,” which explains the basics of genetics and how it is involved in disease; and two higher level courses on genomic technologies in clinical diagnostics: molecular techniques and next generation sequencing. Clinicians, allied health workers, and managers can also access Health Education England funded postgraduate programmes (MSc, PGDip, PGCert) in genomic medicine.

A lot of clinicians are not requesting or receiving any genomic data yet, Tatton-Brown says, “so they don’t realise they need to have the knowledge to interpret it robustly for their patients.”