Outlook | Published:

Diagnosis: Frontiers in blood testing

Nature volume 549, pages S16S18 (28 September 2017) | Download Citation

Technological advances are creating an explosion in possibilities for the blood-based diagnosis of brain injuries, infections and cancers.

The battered sports star of the future won't have to leave the game straight away after receiving a blow to the head. A simple blood test will reveal the severity of the brain injury and tell coaches whether it's safe for the player to return to the game. At least that's the hope — one that's getting closer to reality as scientific advances expand the power of blood testing. And the benefits could extend far beyond the sports field, potentially providing earlier insight into the health of the general public.

Blood testing is expanding its reach beyond conventional blood-based diagnosis. Image: Frank Bienewald/LightRocket via Getty Images

From just a few drops of blood, it may become possible to diagnose not just injuries to the brain but also mental illnesses, cancer (see 'Detecting cancer without a knife') and mysterious infections. Because blood is plentiful and easy to extract, blood tests could replace or precede more invasive, expensive or risky tests. In some cases, blood could deliver diagnoses decades earlier than current methods, and save patients months of uncertainty.

Box 1: Liquid biopsies: Detecting cancer without a knife

One of the most tantalizing applications of the increasing quantity of information that can be discerned from blood is the ability to detect and monitor cancer through traces of tumour DNA circulating in the bloodstream.

In 2016, the US Food and Drug Administration approved the first such 'liquid biopsy' for clinical use. The test is designed to detect two common mutations in the gene EGFR in people with non-small-cell lung cancer. A positive result can change the course of treatment, leading clinicians more quickly to the appropriate therapy, says Geoffrey Oxnard, a thoracic oncologist at the Dana-Farber Cancer Institute in Boston, Massachusetts. The EGFR mutation test is commercially available and already in use worldwide.

Liquid biopsies have transformed the way that Oxnard treats patients. Conventional tissue biopsies are invasive, frequently produce insufficient amounts of tissue for testing, and are capable of sampling just one part of a tumour, even though the genetic composition can vary throughout. It can also take weeks to get results. But liquid biopsies can quickly produce information about the entirety of a tumour. “Now, when you have a patient with advanced lung cancer who is a non-smoker, you can get a blood test back two days later and start targeted therapy,” Oxnard says. In 2016, Oxnard's team found that a blood test works just as well as a tissue-based biopsy for identifying people with one of a spectrum of EGFR mutations, although a negative result still requires a standard biopsy as a follow-up6. And because taking blood is easy, liquid biopsies enable the repeated monitoring of patients, which can reveal genetic changes and the emergence of drug resistance in cancers.

The feasibility of liquid biopsies got a boost in 2015 from the coincidental discovery of metatstatic cancers through cell-free tumour DNA detected in pregnant women whose blood was being examined for fetal DNA. The findings accelerated interest in the idea that blood testing could identify cases of cancer. Rapid next-generation sequencing technologies have made it increasingly possible to measure tiny amounts of circulating tumour DNA, which can make up just 0.001% of DNA in the bloodstream, says Mark Roschewski, a lymphoma expert at the US National Cancer Institute in Bethesda, Maryland.

The long-term goal is to provide initial diagnoses through liquid biopsies. By the time a tumour measures 1 centimetre across on a computed tomography scan, it contains about 1 billion cancerous cells, suggesting the potential for a much earlier diagnosis when such cells or their DNA enter the bloodstream at detectable levels. But for now, liquid biopsies show most promise for understanding the genomics of previously diagnosed cancers, in part because there is more tumour DNA in the blood to work with. Researchers working on the UK-based TRACERx study7 in 2017, for example, used a blood test to pick up early signs of relapse in patients following surgery for non-small-cell lung cancer, enabling the detection of recurrence up to about 1 year earlier than with conventional computed tomography imaging.

The technology is ripe with possibilities, some more distant than others. “What we need is assays that are even better at finding vanishingly low levels of circulating tumour DNA,” Oxnard says, which would open up the ability to diagnose emerging or recurring cancers at the earliest stages — as long as researchers can work out how to apply those results to patient care. “That's the future. It's still very early, but the promise is very real.”

“It has been known for many decades that a lot of information that might help us clinicians make decisions is available in blood,” says Mark Roschewski, a haematologist and lymphoma expert at the US National Cancer Institute in Bethesda, Maryland. “The difficulty has been trying to get that information out — and to try to do it in a way that's accurate, reproducible and timely.”

A fresh slew of blood testing is gaining steam as our understanding of how diseases work grows. Technological advances that allow researchers to scan a sample of blood for many biological indicators of conditions, or biomarkers, simultaneously are also providing a considerable boost. And the discovery of new biomarkers, in turn, is suggesting innovative treatments and enticing pharmaceutical companies to enter a rapidly growing market.

However, as blood tests become more accurate and more comprehensive, concerns surrounding the associated influx of health-related data are emerging. Tests that predict the future occurrence of disease in healthy people are particularly loaded, says H. Gilbert Welch, an academic physician at the Dartmouth Institute for Health Policy and Clinical Practice in Lebanon, New Hampshire, and author of Less Medicine, More Health (Beacon, 2015). “This is the information age, and we think more data will always provide clear answers,” he notes. “But it's really important to distinguish data from useful knowledge.”

Probing the brain

One of the most important developments to drive the swift expansion in possibilities for blood testing is an improved understanding of our immune system's response to various assaults through the release of substances into the bloodstream, together with an increased ability to detect those substances.

“We are starting to realize that the immune system is much more evolved and sensitive than we ever imagined,” says Purvesh Khatri, a computational immunologist at Stanford University in California. “That's why blood is becoming this window into disease.” And it's not just immune-system molecules that can signal assault. An injured nervous system releases proteins that are providing a window into brain science, says Jessica Gill, a neuroscientist at the US National Institutes of Health in Bethesda, who is working on a blood test that could diagnose traumatic brain injury. Her research began with a frequent dilemma: about one-third of US military personnel who go on combat operations receive at least one such injury, but by the time they see a doctor, it can be hard to distinguish the mental symptoms of brain injuries from those of anxiety attacks or post-traumatic stress disorder. Magnetic resonance imaging (MRI) can diagnose brain injuries, but is only definitive when injuries are extreme. “Even when you have an MRI scan, if you have five radiologists, you'll never get agreement unless it's severe,” Gill says. “Most traumatic brain injuries or concussions are really mild — and it's hard to see those with current neuroimaging methods.”

A Simoa single-molecule array. Image: Quanterix

In the search for a better diagnostic biomarker, Gill has zeroed in on a protein called tau, which accumulates in blood at extremely low levels after a brain injury. To see whether tau could also indicate mild or older brain injuries, Gill turned to a technology that uses weakly magnetic beads attached to antibodies to bind to proteins. The accompanying automated analysis system — called Simoa, and developed by health-care company Quanterix in Lexington, Massachusetts — is 1,000 times more sensitive than previous antibody-based assays, and able to detect proteins such as tau at concentrations as low as 0.02 picograms per millilitre of blood, according to Quanterix. “When we're looking at tau, we're looking at something like grains of sand in an Olympic-sized swimming pool, but the grains of sand actually mean something,” Gill says. “We weren't able to do this five years ago.”

In 2015, Gills's team used Simoa to detect increased levels of tau in military personnel with a history of multiple traumatic brain injuries and symptoms of post-concussive complications1. Along with other studies, she says, the results suggest that degeneration of the brain continues for years after repeated injury. Blood tests for tau or other biomarkers could therefore eventually identify the patients in most need of follow-up testing and care. And if the aggregation of tau leads to chronic symptoms after traumatic brain injury, such research might lead to interventions that would target tau.

Early diagnosis of Alzheimer's disease is another high-profile target for blood testing. Positron-emission tomography scans of the brain show that the peptide amyloid-β begins to accumulate up to two decades before the appearance of symptoms of the disease. As researchers search for biomarkers that follow the same pattern, studies have also linked brain injuries and type 2 diabetes with an increased risk of Alzheimer's disease, offering clues about the kinds of biomarkers to search for.

Thanks to assays that can rapidly assess blood for many biomarkers at once — including tau, which has also been linked to Alzheimer's disease — scientists have identified about 100 molecules that might work as diagnostic markers for Alzheimer's disease, says Ralph Martins, a neurobiologist at Edith Cowan University in Perth, Australia. Right now, says Martins, about 20 of them look promising. When that list is pared down to eight, he contends, a blood test for Alzheimer's disease will be in striking range. “I don't think we've got the best ones yet,” he says, adding that a blood test would be far cheaper and easier than scanning for amyloid-β, which can cost thousands of dollars. “The benefits are apparent, and they're huge.”

Blood and mind

Mental health also stands to gain from the coming era of blood testing, says Carmine Pariante, a biological psychiatrist at King's College London who is working on a blood test to funnel people with depression towards personalized treatment. The idea emerged from evidence gathered over the past decade that links depression with inflammation, especially in the one-third of patients who don't show an improvement when taking conventional antidepressant medications.

In a 2016 study2, Pariante and his colleagues searched published data to identify two biomarkers of inflammation (interleukin-1β and macrophage migration inhibitory factor) in blood that, when present above a certain level, accurately identified non-responders to such medications. Before researchers can develop a predictive test, Pariante says, they still have to better understand what particular levels of biomarkers mean, and how individuals vary, especially for the large number of patients with intermediate levels of biomarkers who show a mixed response to standard medications. However, Pariante's findings suggest that blood testing could eventually prevent months of distress as people search for effective treatments.

“At the moment, if you or I visit a primary care physician or psychiatrist to get help, the doctor will prescribe one of many antidepressants that are all, more or less, equally effective, but there is no way of deciding which is more likely to work for me or you,” says Pariante, whose research group is one of several that are working on ways to test blood for further biomarkers of depression. “You will spend six months depressed and taking medication with all the side effects and nothing good.”

Studies such as these, which highlight depression's biochemical underpinnings, are prompting pharmaceutical companies to renew their interest in mental health, Pariante says — and word has already filtered out to those affected by these disorders. “Patients that write to me say how much they recognize that mental-health problems are not just something they have in their mind, but also in their bodies,” he says. “This really demonstrates that we should not even discuss mental and physical health as separate concepts.”

Infection detection

As scientists get better at deciphering the immune system's fine-tuned responses to attack, they are also establishing systems for data sharing that, when combined with blood testing, will help to investigate infections. These developments could save lives and reduce the unnecessary use of antibiotics, says Khatri. His group has been mining public databases, including those maintained by the National Institutes of Health and the European Bioinformatics Institute in Cambridge, UK, to unearth patterns of gene expression that correlate with causes of infection. Those patterns can then be used as a warning sign to identify people who may become ill or contagious.

Among their findings, in 2015, the group identified 396 genes that showed distinctive patterns of activity in people with either a bacterial or viral respiratory infection3. And they also found a unique signature that distinguished influenza from other viral illnesses. A follow-up study4 identified seven genes that distinguished bacterial infections from viral infections, which were tested for accuracy in a group of 96 critically ill children. The team has also identified a signature of three genes that discriminates between people according to whether they have active tuberculosis, a latent version of the disease, another disease or nothing at all. The test is extremely accurate at identifying people without the disease, Khatri says, unlike other tests for tuberculosis, which detect bacterial DNA in mucus but often miss it in young children and patients who can't spit up enough phlegm for testing.

In other studies, Khatri's team has uncovered gene-expression patterns that correlate with and can distinguish the parasitic disease malaria from bacterial and viral infections. And they have developed an 11-gene test that can separate people with sepsis (a full-body, and often fatal, form of inflammation caused by a bacterial infection) from those with an inflammatory response to a non-infectious cause, such as a blood clot or traumatic injury. The test was able to identify those who would develop signs of infection up to three days before symptoms appeared, enabling sepsis to be diagnosed five days earlier than is possible at present using laboratory-culture methods. Because patterns of gene expression often change before symptoms appear, Khatri says that such blood tests could help to battle the spread of diseases by predicting who will respond to vaccines or by identifying those who are infectious but not yet unwell. In their 2015 study3, the team used blood testing to recognize people who, despite being asymptomatic, were shedding the influenza virus, as well as those who had apparent symptoms but were not infected. Current rapid blood-testing technologies are at most only 70% accurate, Khatri says, which means that if they were used on the roughly 25 million people who get influenza in the United States each year, they would miss about 7.5 million.

The growing ability to diagnose infections in blood would not be possible without widespread data sharing, says Khatri. More than 1 million 'catalogues' of gene expression, known as transcriptome profiles, in humans are now freely available for investigation. Linking innovative techniques for analysing blood to these massive databases could provide great benefit. “Blood tests used appropriately and in the right context are going to empower precision medicine,” he says.

As the list of health issues with biomarkers that are known to circulate in the blood expands to include conditions such as osteoarthritis, autism and heart disease, some scientists are looking to develop comprehensive screening tools that can spot a wide variety of conditions using blood before the symptoms set in. In one of the earliest demonstrations of the clinical possibilities, in 2014, researchers from Arizona State University in Tempe and from San Diego, California-based company NextVal described a platform that uses microarrays to simultaneously scan less than 1 microlitre of blood for as many as 330,000 peptides that can be used to identify circulating antibodies5. Using blood drawn from people with known diagnoses, the method was able to identify six types of cancer and six infectious diseases, including Lyme disease and whooping cough.

In other work, researchers at Stanford University School of Medicine are using a technique called mass cytometry, which tags antibodies with heavy metal ions and detects them by mass spectrometry, to investigate cells of the immune system, allowing the team to use more antibody probes at once than is possible with other methods. For now, the expensive technique is focused on discovering biomarkers, but that could change. “Our goal is to eventually design a test that tells you how well your immune system is doing: maybe your cancer risk, maybe your cardiovascular disease risk, or other disease risk based on your immune profile,” says immunologist Holden Maeker at Stanford University. “It comes down to inflammation and immunity being involved in a lot more diseases than we maybe initially appreciated,” he says.

Cost and speed remain obstacles to the development of practical blood tests, and the fate of health-technology company Theranos, based in Palo Alto, suggests the need to caution against overhyping. Despite heavy investment, its technology failed to deliver. But even as biomarker-detection technologies improve, such tests will never be definitive, says Welch, partly because they often look for differences between groups of people who are either healthy or very unwell — even though most people inhabit a vast grey area, in which lines are hard to draw. And when blood tests are able to offer useful information, they don't necessarily reduce the need for more invasive follow-up testing.

The prostate-specific antigen blood test for prostate cancer is a perfect cautionary tale, says Welch. Early on in its use, abnormal results led many men down a path of extra testing, costs and worry, even as researchers shifted their definition of how high was too high a threshold for the antigen. Efforts to anticipate downstream health problems raise “the potential that problems enter people's lives earlier in life,” Welch says. “The truth is, it's really hard to predict the future.”

This article is part of the Blood Outlook, an editorially independent supplement produced with the financial support of a third party. About this content.


  1. 1.

    et al. JAMA Neurol. 72, 1109–1116 (2015).

  2. 2.

    et al. Int. J. Neuropsychopharmacol. 19, pyw045 (2016).

  3. 3.

    et al. Immunity 43, 1199–1211 (2015).

  4. 4.

    , & Sci. Transl. Med. 8, 346ra91 (2016).

  5. 5.

    et al. Nature Commun. 5, 4785 (2014).

  6. 6.

    et al. J. Clin. Oncol. 34, 3375–3382 (2016).

  7. 7.

    et al. Nature 545, 446–451 (2017).

Download references

Author information


  1. Emily Sohn is a freelance journalist in Minneapolis, Minnesota.

    • Emily Sohn


  1. Search for Emily Sohn in:

About this article

Publication history





By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing