Physicist Peter Kuhn views himself and his colleagues primarily as inventors and problem solvers. But the problem that he and many other physical scientists are trying to solve has plagued biologists and oncologists for decades: how to detect cancer early and stop it from spreading.

Nanoscale density variations in cancer cells (red colour, top) compared with normal cells (bottom). Credit: BACKMAN LABORATORY, NORTHWESTERN UNIVERSITY

Physical scientists have long played an integral role in cancer diagnostics. Magnetic resonance imaging, computed tomography, and positron-emission tomography scans — crucial imaging technologies for identifying cancer and tracking its growth — were developed by physicists, mathematicians and engineers. Physical scientists continue to improve and combine these technologies, but they are also moving into the role of pathologist, developing tools to help physicians spot cancer in blood and tissue samples. These new technologies, ranging from nanosensors to souped-up microscopes, promise to help find cancer earlier than previously thought possible.

Solid tumours shed cells that enter the bloodstream and travel to other parts of the body, where they can take hold, forming metastases. “This fluid phase of solid tumours is something that's really poorly understood,” says Kuhn, who is based at the Scripps Research Institute in La Jolla, California. One reason is that circulating tumour cells are so few in number: there may be only a handful of cancer cells among the billions of cells in a tube of blood. “It's a rare-event detection problem,” Kuhn says.

To identify cancer cells, Kuhn and his colleagues have developed a technique called fluid biopsy, which can be done using a two-millilitre vial of blood. The researchers eliminate all the red blood cells from the sample and lay what remains on a slide in one “massive monolayer” of 10 million cells, Kuhn says. Then they apply fluorescent stains: a chemical to highlight the cells' nuclei, an antibody that adheres to immune cells, and another antibody that tags epithelial cells. A digital microscope takes approximately 10,000 images of the slide. “Then we start slicing and dicing,” Kuhn says, using the different stains and a computer algorithm to isolate the epithelial cells that lie at the root of many important cancers — including breast, lung, colon, prostate, pancreas and liver cancer.

Once these cancerous suspects have been identified, “you can do everything you do with a regular biopsy”, Kuhn says, such as look for abnormal morphologies. Kuhn's research indicates that the fluid biopsy can detect even more circulating tumour cells than the CellSearch assay, a test approved by the US Food and Drug Administration (FDA) that uses antibodies and magnets to isolate tumour cells.

Kuhn and his colleagues are currently investigating whether the number of circulating tumour cells can predict disease progression or treatment response, but in theory the fluid biopsy could also be useful for cancer diagnosis. The team recently showed that their technique could detect circulating tumour cells in blood samples from patients with non-small-cell lung cancer, even in the early stages of disease1.

Realizing its potential to diagnose cancer will require some significant advances, says Kelly Bethel, a pathologist at Scripps and Kuhn's collaborator. She is seeking grants to use the technique to search for circulating tumour cells in individuals with a high risk of cancer, such as patients with unidentified masses in their lung or pancreas. Such an approach could help physicians distinguish benign growths from malignant masses that need immediate attention.

The current version of the technology is unable to identify where a circulating tumour cell came from, but future versions might be able to provide that information — the researchers are already searching for a liver-specific marker. “All that potential exists,” Bethel says, “it's just not well developed.”

The protein hunter

Sanjiv Gambhir, chair of the radiology department at Stanford University in California, has also been frustrated by physicians' inability to detect cancer at an early stage. A breast-cancer lump the size of a marble can contain as many as 3 billion cells, he says, and has probably already shed cells that have travelled to other parts of the body. Rather than waiting for a lump to grow to such a potentially lethal size, he is working on a way to detect tumours when they are smaller than the head of a pin.

Gambhir and his colleagues have developed a postage-stamp-sized chip that uses magnetic nanoparticles to detect cancer proteins in the blood “at much lower levels than otherwise possible”, Gambhir says2. When researchers place a drop of blood or other bodily fluid onto the chip, antibodies on its sensor strip lock on to any cancer proteins and secure them to the chip. Then they add more cancer-binding antibodies. “It's called a sandwich assay,” Gambhir says. “You've trapped the thing you're looking for between two antibodies.” The next step is to wash away any unbound material and add tiny magnetic nanoparticles that bind to the antibodies. When cancer proteins are present, the magnetic particles change the magnetic field and trip the sensor. The researchers then insert the chip into a separate device that identifies which antibodies have been bound and reveals whether any proteins known to be associated with specific cancers are present.

The chip can detect exceedingly small quantities of protein, a thousand times less than those needed for a standard ELISA assay, the method most commonly used to detect cancer proteins in the blood. The latest version of Gambhir's chip can detect 256 different cancer proteins, improving the test's sensitivity and specificity. But the device is only as good as scientists' understanding of cancer biology, so the chip's development team must work closely with biologists to decide which combinations of proteins are important. “This technology can't work miracles,” Gambhir says. “If you don't know what [proteins] to look for, it can't help you.”

Under the microscope

While researchers such as Gambhir are developing new devices to improve cancer diagnosis, other groups are following the same path as Kuhn, trying to boost the diagnostic potential of a centuries-old device — the microscope. VisionGate, a company based in Phoenix, Arizona, has developed an optical technology called Cell-CT, which creates three-dimensional digital images of cells. Cells are lined up single-file in a small, gel-filled tube; as the tube rotates, the microscope captures images of them from multiple angles. A computer algorithm then turns these pictures into a three-dimensional image that researchers can use to identify structural abnormalities that are associated with cancer.

“You sometimes find that the nuclei look like a beach ball that somebody sat on,” says theoretical physicist Paul Davies of Arizona State University's Center for the Convergence of Physical Science and Cancer Biology in Tempe. Davies and his colleagues, who are testing the technology, recently reported that normal, benign and malignant breast cells from different cell lines displayed distinct morphological characteristics that probably would not have appeared in a conventional two-dimensional image3.

Cell-CT is the only imaging technology able to create a true three-dimensional reconstruction of a cell, says VisionGate's chief executive, Alan Nelson. The rendering is so precise that the system can make hundreds of measurements within the cell to search for 'biosignatures' associated with cancer. “You don't actually need a pathologist. It's all automated,” says electrical engineer Deirdre Meldrum, director of Arizona State University's Center for Biosignatures Discovery Automation. VisionGate is developing the technology as a screening tool for lung cancer, and has found evidence that the system can detect cancerous cells in the sputum of individuals at high risk of developing the disease.

Vadim Backman, a biomedical engineer at Northwestern University in Evanston, Illinois, has developed a microscope designed to detect even subtler signs of disease. “A limitation of [traditional] microscopy is that you can only see structures larger than half a micrometre,” Backman says. But his research suggests that the earliest signs of cancer appear as nanoscale abnormalities.

To evaluate these tiny aberrations, Backman and his colleagues have developed an instrument that combines traditional microscopy with spectroscopy. The partial-wave spectroscopic (PWS) microscope illuminates the cell and analyses the reflected wavelengths. Their research suggests that the epigenetic and genetic changes that precede cancer cause nanoscale variations in the density of the cell, even though the cell appears superficially normal. The greater the variation, or 'disorder strength', the closer the cell is to becoming cancerous.

As well as appearing in the location where a tumour will occur, the abnormalities also affect nearby cells. This suggests, Backman says, that PWS could potentially be a valuable screening tool. A physician might be able to detect signs of lung cancer in cheek cells, which are easy to obtain, for example. Backman and his colleagues recently used PWS to examine rectal cells from more than 100 individuals who were undergoing colonoscopies, and found a strong correlation between cellular disorder at the nanoscale and the risk of colorectal cancer4.

The sooner, the better

Physical scientists hope these tools will help oncologists detect cancer early, before it gains a foothold in the body. The five-year survival rate for lung cancer caught early is about 50% — but catching it early is incredibly difficult, because symptoms typically appear only during advanced stages of disease. Once the cancer has spread, the survival rate drops to just 1%.

Ideally, screening tools and diagnostics would not only be able to detect early signs of cancer, but also differentiate between harmless changes and abnormalities that precede the disease. However, that's a formidable challenge that will keep Kuhn and his inventor colleagues busy for decades to come.