Hospitals rush to get the latest multi-Tesla MRI machines, physicians eagerly prescribe the latest biotechnology-derived medicines, and medical laboratories happily deploy cutting-edge genomic tests on a daily basis.
Are you neo-Amish? It’s the hot new buzzword in Silicon Valley, describing technophiles who’ve eagerly adopted almost every new gadget, but then, incongruously, decided to reject one. The neo- Amish include the computer scientist who spends his days writing innovative software but can’t be reached by e-mail, the serial entrepreneur who refuses to own a cell phone, and the Web guru who doesn’t have a blog.Medicine, though, is the original homeland of the neo-Amish. Hospitals rush to get the latest multi-Tesla MRI machines, physicians eagerly prescribe the latest biotechnology-derived medicines, and medical laboratories happily deploy cutting-edge genomic tests on a daily basis. Meanwhile, many clinicians still keep patient charts with paper and pen, communicate with hopelessly outdated pagers, and often transfer information by entering it manually from one computer system to another.Regulatory and financial barriers may keep some of these antiquated approaches active for a while longer, but gadget-loving doctors can take comfort in a new crop of technologies that will transform other aspects of medicine. That’s especially true in oncology, where vibrant collaborations between clinicians, researchers, and engineers are producing intriguing new tools. Ranging from electronic resources that are already available to future diagnostic and treatment systems, a sampling of these technologies shows how they could change your practice dramatically.
The pharmacist who never sleeps
At the University of Colorado medical center in Denver, CO, the oncology pharmacy of the future is already up and running. Sealed inside a containment chamber, a robot makes a series of precise movements, measuring and mixing chemotherapeutic agents and loading them into syringes or IV bags for delivery to patients. The robot, called CytoCare, was designed to overcome a slew of problems that have plagued chemotherapy administration for years.
“The first element was sort of the patient priority, and making sure that we are delivering...the right dose to the right patient,” says Jack Risenhoover, Senior Partner for North American operations at Health Robotics, the system’s manufacturer. Besides using both bar codes and computer vision technology to ensure that the medications are correct, CytoCare also weighs each dose and the fi nal mixture to be sure they’ve been measured accurately.
That alone provides a significant improvement over human pharmacists who measure doses with plastic syringes. “The variation just of the syringe marks can be plus or minus 5%, so CytoCare is usually at least 50% more accurate...than what you could do manually if the only thing that you looked at was the variation in the black lines on the syringe,” says Risenhoover. In addition, the robot is immune to the errors that can creep into human judgment, such as technicians’ natural tendency to underdose toxic medicines.
Pharmacists need not fear that robots will take away desirable jobs, however. Indeed, recent research has highlighted the many hazards that chemotherapy poses for pharmacy technicians— from exposure to the cytotoxic drugs themselves to repetitive strain injuries from drawing each dose into and out of a syringe.
The contamination doesn’t stop at the pharmacy doors, either. “Aerosols happen, aspirations happen, errors happen, leaks happen, and it doesn’t take very many of them for you to have contaminants in a variety of places, from the laminar fl ow hood where the technician performs that initial survey...to the fl oors where the syringes and bags are administered,” says Risenhoover. With the robotic system, pharmacists simply insert the sealed, bar-coded bottles of medicine through an air lock, and collect the sealed, fi nished bags or syringes for each patient’s dose from the same portal a few minutes later.
A complete system costs about $1 million, but the company estimates that it will save money in the long run by limiting errors and freeing pharmacists for other tasks. Because the robot documents each measurement and movement meticulously, it could also provide important new data on drugs. “It gives you a much more fertile, accurate field of information for future research to be able to evaluate which drugs had what kind of consequences downstream when you have an electronic medical record of all of the compounding activities that took place,” says Risenhoover.
Technophile oncologists don’t need to wait for a new robot if they want to be at the cutting edge of patient management, however. Through the National Cancer Institute’s (NCI) new Cancer Biomedical Informatics Grid (caBIG), anyone with an Internet connection can now tap into a trove of useful data about cancer.
Scientific and medical databases are certainly not a new idea, but most of these tools have amassed so much raw data, in so many diff erent formats, that they are virtually unusable by anyone but hardcore bioinformaticians. Over the past few years, caBIG has been connecting these disparate databases, standardizing their content, and equipping the whole interlinked system with user interfaces that working physicians can understand.
For example, caBIG recently released the fi rst version of a clinical trial software suite. “This is different from typical clinical trial software that’s really designed for the sponsor, for drug companies, and so on to operate,” says John Speakman, Associate Director of Clinical Trials Products at the NCI in Bethesda, MD.
Instead of simply collecting data for the drug company, the new programs track and unify the information that oncologists and hospitals actually need, such as which patients are participating in multiple trials, when their chemotherapy appointments are scheduled, and which adverse events they’ve experienced. “Typically, what you do with adverse events is you just send a report off to the sponsor or the FDA or whoever and you don’t keep local track of it,” says Speakman.
The caBIG suite helps prepare those reports, but also tracks non-serious side effects, makes sure the patients’ electronic medical records mention the problem, and modifi es the treatment schedule automatically. “If the patient experiences an adverse event, chances are that their schedule of treatment will have to be modifi ed, and so you don’t actually have to keep track of these things yourself,” says Speakman.
Other software from caBIG, all of which is free and open source, includes an application that can transfer data from non-standard clinical laboratory systems into a hospital’s central database, eliminating manual data entry, and a program that tackles the onerous but essential regulatory paperwork associated with clinical trials. Another system, called caMATCH, provides a portal for oncologists and patients to fi nd clinical trials that might benefi t them.
Most of the current users of caBIG are physician-researchers at large medical centers, but that could change as the system develops. Indeed, one of the NCI’s goals with the program is to make it easier to manage clinical research so that any oncologist can do it. “What we want to do is lower the barriers ... toward regular physician practices, community medical centers, and so on participating in clinical trials,” says Speakman.
New biomarkers, this time for real
Unsurprisingly, the NCI is also working on several other cancer-fi ghting technologies. Th e fi rst tools practicing oncologists are likely to see from these tightly linked eff orts are new biomarkers. In recent years, numerous projects have promised to fi nd new cancer biomarkers, but in this case, they might actually materialize.
The difference is that the NCI is focusing on the core problem in the cancer biomarker fi eld, which can be summarized in three words: validation, validation, validation. Researchers seem to report a new marker for diagnosing or staging cancer every day, but further studies often reveal these tests to be unreliable or impractical. Now, the NCI has stepped in and built a complete pipeline for biomarker discovery and validation, and some physicians are already taking advantage of it.
One of the effort’s most popular programs right now is the rapidly expanding Cancer Genome Atlas, which aims to collect the signifi cant DNA mutations that appear in all major types of cancer. So far, the project has amassed a large data set on glioblastoma, and data on ovarian cancer and lung cancer are now pouring in as well. “We do have clinicians who are starting to look at the Atlas already in terms of how they’re classifying their patients. You might use these data to try to stratify patients into treatment groups,” says Anna Barker, PhD,, Deputy Director of the NCI.
The institute also wants doctors to start giving something back to the program, in the form of more carefully collected clinical specimens. “This is an area that we’ve had to work on a great deal, because there were no standards for collecting these tissues, [and] we want to be sure that we’re measuring in these tissues the changes in genes related to disease, not related to something like anesthesia,” says Barker. Indeed, the Atlas’s initial focus on glioblastoma was driven partly by a shortage of wellcharacterized tissues for other tumor types, so getting oncologists to collect better samples is critical for the project’s future.
To make the Atlas more accessible, NCI is collaborating with the FDA and other agencies to ensure that new biomarkers and sample collection procedures will be both approvable and billable. “It’s going to be expensive... to do these kinds of things well, you’re going to have to have very highly characterized, clinically annotated samples, which is not something that people have been accustomed to doing,” says Barker.
A PET for your MRI
“Expensive” will probably also describe another breakthrough technology now on its way into the clinic: the combined PET/ MRI machine. While PET/CT systems have become standard equipment in recent years, combining the functional imaging of PET with the fi ne-grained structural imaging of MRI has remained more a dream than a reality.
“Often, in order to interpret the signal you get from the PET, you really need to understand exactly where that signal’s emanating from anatomically. That’s particularly true in cancer, because you know if you’re looking at metastasis, exactly where those metastases are has a big impact on how you treat the patient and the patient’s outcome as well,” says Simon Cherry, PhD, Chair of Biomedical Engineering at the University of California in Davis, CA. The higher resolution of MRI compared to CT could be especially useful when tumors are buried in soft tissue, such as in the pelvis or the brain.
Getting PET and MRI into a single system, however, has been hard. The MRI machine’s massive magnetic fi elds and powerful radio frequency pulses will obliterate a standard PET scanner, which, in turn, generates more than enough electromagnetic noise to befuddle the MRI. By carefully redesigning both systems, however, Cherry and his colleagues recently managed to install a PET scanner inside an MRI machine, creating a system that performs both scans simultaneously.
That design contrasts markedly with the layout of a typical PET/CT scanner, which runs the two scans in tandem rather than simultaneously. For Cherry, research needs drove that decision: “If I wanted to study both the permeability of the brain to [a] drug and then also see what happens, where the drug actually goes inside the brain, I could measure permeability with MRI, and I could measure the drug itself, radiolabeled, with PET and look at both of those things at exactly the same time.” He adds, “if I had to take the scans, 10, 20 minutes apart, I wouldn’t be able to do that.”
For hospitals, the simultaneous scanning should improve patient throughput, and might also enable new types of data analysis. Physicians won’t have to wait long to fi nd out if the system lives up to its promise. Siemens (Munich, Germany) has already delivered a prototype of a clinical PET/MRI system to Massachusetts General Hospital in Boston, MA.
Whatever modern imaging technique physicians use, the final product is inevitably a kind of digital photograph. In order to shoot those pictures faster and for less money, oncologists may soon be using a somewhat counterintuitive technology: the one-pixel camera. “This doesn’t sound so exciting at first. Everyone wants megapixels; this has one pixel, but in terms of medical imaging, pixels are really expensive,” says Kevin Kelly, PhD, Assistant Professor of Electrical Engineering at Rice University in Houston, TX.
The problem is that medical imaging relies on electromagnetic signals well outside the visible light range, where the cheap silicon detector chips used in consumer cameras become useless. Instead, medical imaging systems must use more exotic detectors, giving clinicians Hobson’s choice between sub-optimal resolution and an unaff ordable price.
Rather than try to come up with a cheaper high-resolution detector, a goal that has so far eluded researchers, Kelly and his colleagues pull more information out of each pixel. “Our single-pixel camera technology is something that will allow you to take maybe 10,000 measurements and build up a 1-megapixel image, and it does this with some really cool mathematics,” says Kelly.
In a typical experiment, the researchers take an incoming electromagnetic image, then apply a known modulation signal to it and feed it through the single-pixel camera. “The reason you don’t normally want to do that is, if there’s light hitting your eye or light hitting your camera sensor, if you threw all the light together you wouldn’t understand the picture, but it turns out through special mathematics and controlling the modulation, we can easily understand the picture and pull back out a visual image,” says Kelly.
The process is superfi cially similar to image compression techniques such as JPEG, which shrinks large digital pictures into fewer bits and then expands them again later. However, the mathematical algorithm Kelly uses can actually reconstruct the original image perfectly, with no loss of fi delity.
When the one-pixel system enters clinical use, which Kelly predicts will be within fi ve years, doctors could face a much happier choice in imaging systems. A relatively cheap one-pixel setup could enable smaller clinics and practices to aff ord sophisticated imaging technologies, while higherpriced units might incorporate larger detectors to accelerate the pace of care at big hospitals. “You could think about it as higher throughput of patients through there, or you could think about it as cheaper,” says Kelly. For CT and other radioimaging techniques, faster scans could also reduce patients’ exposure to radiation. The one-pixel camera—and all of the other new medical technologies now in development—still have to prove themselves both clinically and economically before becoming standard tools. Th ere’s little doubt, though, that at least a few of them will change your practice, if they haven’t already. Now, if only we could do something about those darn pagers.
Alan Dove, PhD, is a freelance healthcare and science writer.