Illumina has created a splash by announcing the formation of a new company, GRAIL, to develop a ‘pan-cancer screening test’ using patient blood samples. The aim will be to use Illumina's sequencing firepower and expertise to develop tests to analyse circulating nucleic acids – DNA and possibly also RNA – in the blood, with the goal of detecting circulating tumour (ct) DNA in otherwise asymptomatic patients.

The basic premise seems reasonable – cancers detected early in their development are less likely to have metastasised and are therefore easier to treat successfully. However, there are many technical challenges that will have to be managed and important questions surrounding the clinical validity and utility of tests. In addition, concerns that screening could lead to over-diagnosis and over-treatment should be addressed.

There is currently little (if any) evidence that measuring ctDNA in blood can diagnose asymptomatic patients. One company in the United States, Pathway Genomics, began marketing a test in September last year as a ‘cancer stethoscope’ for those patients determined to be at high risk of developing cancer (for example, smoking or family history). They received an untitled letter from the US Food and Drug Administration (FDA) for their efforts, effectively a warning that although they would not be subject to official action by the FDA, they should change their marketing approach, given the lack of evidence that these tests are clinically relevant.

Clinical utility and implications

Among the questions that the GRAIL researchers will have to contend with are:

  • Is the ctDNA test an accurate reflection of clinical reality? What does the presence of ctDNA in the blood represent, particularly in an asymptomatic patient? Does it represent a tumour that could cause the patient harm, or a slow growing malignancy that is unlikely to need treatment? What if the presence of ctDNA cannot be explained given the information available?
  • Threshold, sensitivity and specificity of tests. How much earlier can ctDNA tests detect a cancer compared with conventional methods? How much of a clinical advantage will this confer to the patient? Current ctDNA tests/assays have been developed for use in patients whose cancers have already been diagnosed using other methods. The tests need to be specific enough to determine the driver mutations in the cancer as well as its tissue of origin, complicated by the fact that certain genes are implicated in more than one cancer e.g. KRAS mutations are found in some pancreatic and colon cancers, less commonly in other cancers. Understanding this information will require that extensive data on the genetics of cancers are freely available.
  • Are other technologies sensitive enough to confirm a ctDNA diagnosis? Tumours diagnosed via ctDNA tests in asymptomatic patients are likely to be small if discovered early. It is likely that some of these tumours will be hard to detect (or undetectable) using currently available scanning technologies, making clinical follow up difficult in some cases.
  • Potential burden on the health system. There is a danger that over-diagnosis and over-treatment could lead to a burden on the health system, through spending money on costly further investigations and interventions, to the psychological impact on patients. A balance will need to be reached between managing the benefits of early diagnosis and treatment versus the risks of over-treatment.
  • What will the effect be on survival rates? There is still much debate about the benefits of screening for different forms of cancer. For example, while screening for breast cancer using mammography detects some cancers at an earlier stage, many studies have demonstrated only a modest decline in rates of advanced cancers. For example, extensive clinical trials will be required to determine if patients whose cancers are diagnosed earlier using this technology have better survival outcomes than patients who are diagnosed via conventional methods.
  • Cost. For this type of technology to be used in NHS cancer services, extensive cost-benefit analyses will be required to determine if this approach saves money in the long run. Developing tests that work is one thing, developing cheap tests that work is quite another.

The bottom line

In short, this is going to take a lot of work to take the tests to clinical practice, particularly validating techniques in thousands of patients, including clinical follow up. However, although the challenges are extensive, they are not insurmountable. It will be interesting to see how the expertise and clout of Illumina and GRAIL rise to the challenge – if they can deliver, the changes to the way cancer is diagnosed and managed will be profound.