BioWhat?

BIOmedical Data  

by Mike King

Why a science you’ve never heard of is changing the way medicine is practiced.

You know about the World Wide Web.

Think instead about a new generation of the web specifically designed to store detailed medical information from patient records supplied by the best research institutions around the world. Next, think about a Google-like search engine capable of analyzing and comparing all that patient data, as well as radiologic images, pathology slides, genetic information, and even physician notes.

Finally—thanks to some of the most brilliant minds in high-end computer technology—put the whole data collection, retrieval, and dissemination process on what could be described as hyper-drive so that any physician can use it in deciding how best to treat an individual patient.

Now you’re getting close to understanding the ever-expanding field of biomedical informatics.

From helping classify the smallest lesions inside the human brain to reducing the rate of re-admissions for patients leaving the hospital, biomedical informatics is poised to dramatically alter medicine and public health for decades to come.

Researchers, clinicians, administrators, and health officials have barely cracked open the toolbox available to them by expanding the use of computer-assisted biologic, medical, behavioral, and health data. New information resulting from the use of biomedical informatics is coming, at times, at a dizzying pace. And with it, renewed hope for real progress in diagnostic and treatment methods for stubborn killers like brain, ovarian, and lung cancer.

Creating more than electronic medical records

Much of the media attention about biomedical informatics in recent years has focused on electronic medical records systems. Besides reducing medical errors, streamlining the sharing of information among physicians, improving discharge planning at hospitals, and other improvements in quality of care, computer platforms are being built to allow patients to manage their own health care data through PCHs (personally-controlled health records). Still, this barely scratches the surface of what could be achieved with the use of biomedical information technology.

Beyond the work that already has begun with cancer research, biomedical informatics has potential of advancing the treatment for more common conditions (such as asthma and heart disease) by better understanding genetic, environmental, and behavioral triggers for disease.

Deeply connected to the expanding field is the Emory University Center for Comprehensive Informatics. Through a multidisciplinary team approach, the center uses computer technology to analyze and interpret vast sums of data generated within Emory’s research and clinical community, as well as research data collected at cooperating institutions around the country. Besides advancing work in clinical medicine, the center works with the private sector to develop new and innovative platforms to tap into the field.

Leading the effort is Joel Saltz, an Emory pathologist and computational scientist, who is a pioneer in the field and a Georgia Research Alliance Eminent Scholar. Besides years of study and training, he has a deep personal interest in both math and physics. He was attracted to parallel computing because he saw the potential of computational science to help solve biologic problems faster than with use of conventional resources.

“The intersection of medicine and informatics always intrigued me,” he says, “especially if we can make a difference in medical diagnosis and outcomes.”

Changing how clinical research is conducted

Not that long ago, cancer researchers kept most of their records on paper and shared them with fellow researchers by sending tissue samples through the mail and notes via facsimile. Clinical researchers relied on local pathology labs to help them determine treatment plans for patients.

Even after the web was up and running, clinicians relied heavily on medical journals, scientific meetings, and other methods of dissemination of research to make critical decisions about how to help patients.

Modern medicine no longer affords such quaint, and slow, research protocols.

BIOmedical DATA

The Blue Ridge Academic Health Group, which studies health care initiatives and includes several Emory experts, estimates that a physician would need to read the equivalent of 16 scientific articles per day and review dozens of clinical trials to stay abreast of the latest developments. Even then, it would take an exceptional human brain to store and readily retrieve the information gleaned from those articles to put it to use in diagnosis and treatment.

In reality, determining what that data might mean takes the information processing capabilities of a computer connected to specific data sets containing all that information and—most important—expert help in analyzing and interpreting the data.

Emory oncologist Otis Brawley, chief medical officer of the American Cancer Society, believes that bioinformatics—especially the use of molecular analysis to identify what is going on inside tumor cells—“is clearly where oncology is going in the years to come.”

For more than a century, cancer was defined largely by what a pathologist observed under a microscope, Brawley says. “But as we progressed to the molecular age, we’ve seen some things in the biopsies of cancer patients that behave like cancer and some things that don’t. That’s allowing us to redefine cancer, how and when to treat it, and how we might prevent it from recurring.”

Mining data from the human genome

Cancer informatics got a huge boost in 2003 on completion of the Human Genome Project. The U.S. Department of Energy and the NIH spearheaded the effort to map and identify the human genome and then placed all the assembled information in databases for medical use. That opened the door to link clinical findings and molecular analysis for the first time on a large-scale basis.

Here’s a hypothetical example of how biomedical informatics could alter an individual patient’s treatment and become the basis for recommendations on how to prevent disease: A 45-year-old patient with colon cancer develops a small tumor less than a year after having polyps removed. His doctor wants to know how many men between 40 and 50 who have had polyps removed are found to have small tumors with the same pathology within 12 months after a colonoscopy.

Using a database of tissue samples and fast, detailed molecular analysis of the tumors could help determine when and how colon polyps of the kind his patient has will eventually become cancerous. That, in turn, could lead to targeted therapies for his patient and, as more data is accumulated, to prevention recommendations for men in their 40s who have a family history of colon polyps.

Saltz believes that virtually every use of biomedical informatics will, in some form, serve to support the cause of better public health—by improving treatment or helping control costs by providing more information about what works and what doesn’t.

Emory’s connection to the Cancer Informatics Grid

Saltz came to Emory in 2008 after groundbreaking work at Ohio State University, where his research group helped develop caGrid, the programming infrastructure that enables information and analytical resources to be efficiently and securely shared among cancer researchers. The open-source information network that caGrid supports is called caBIG (Cancer Biomedical Informatics Grid), a major initiative of the NCI.

Saltz likes to describe the clinical role of biomedical informatics as “helping remove the subjectivity” in dealing with disease by being able to hone in on what types of triggers are causing it, how severe those triggers are, and how to classify them to make clinical care more precise.

Moreover, he says, biomedical informatics opens the door to new categories of disease by using high-end computing, digital imaging techniques, and other tools to identify medical problems before they can be seen, or even predicted, by conventional clinical observation. That’s also a good description for the largest—and perhaps the most ambitious—project using biomedical informatics at Emory.

Backed by a $2.2 million grant, the Center for Comprehensive Informatics was chosen in 2009 as one of six centers for “in silico” (via computer) research on cancer. The first round of the three-year grant has concentrated on molecular, pathology, and brain tumor data obtained through the Cancer Genome Atlas (TCGA) and other sources residing on the caGrid. Emory pathologist Daniel Brat is the principal investigator.

The research uses rich, molecular data sets and links them to digitized pathologic slides and annotated neurologic images. It also correlates detailed information about patient treatments and results. This approach allows brain tumor research to be conducted on a much larger scale and at a much deeper level than ever before, Brat says. Similar work is going on at other centers, using TCGA data available for lung and ovarian tumors—two other forms of cancer that have proven to be stubbornly resistant to conventional therapy.

“The advances in digital pathology alone have been remarkable in just the past few years,” Brat says. “We can scan in a slide at high resolution and use computer-based algorithms to extract patterns and features that the human eye cannot recognize. We already have developed algorithms that show us which regions of the brain are normal and abnormal and others that can discriminate between types of brain tumors. Eventually, this will help us better classify brain tumors and predict their progression.”

Using biomedical informatics to learn more about brain tumors

Patients who agree to participate in the research allow tissue samples, radiographic images, clinical notes, and other details of their diagnosis and treatment to be shared with the cooperating institutions, which in turn agree to protect patients’ identity. Names and other information that would personally identify the patient do not become a part of the record.

The first round of brain tumor research was conducted on glioblastomas, the most common and most malignant of brain tumors. (Sen. Edward Kennedy was diagnosed with this form of brain cancer in 2008 and died of it the following year.)

Using pathology slides and detailed patient information collected at a dozen institutions nationwide, researchers at Emory discovered—in the space of a few months—what otherwise would have taken years to find out about a range of genetic mutations in some glioblastomas.

Brat believes that correlating this information may allow researchers to target therapies for specific tumor types to keep them from triggering the genetic “on/off” switch that spreads the cancer so rapidly and leads to death. Similarly, researchers may find that some forms of therapy cause a hyper-mutation of genes and therefore should be avoided in some patients.

More recently, TCGA has expanded its efforts beyond the initial pilot study of glioblastoma, ovarian, and lung cancer to 20 additional forms of cancer. Brat is leading the national effort on lower-grade gliomas, which should advance the understanding of these diseases through Emory’s in silico study too.

One of the goals of this research is to use the same kind of data sets to determine if there is a common genetic mutation that turns some lower-grade gliomas into the more deadly glioblastomas. Brat and researchers at four other centers will be looking for the molecular pathways that control growth of the smaller tumors in hopes of one day generating therapies to prevent the spread of cancer cells.

Compiling and disseminating all this information takes time and demands an unprecedented degree of cooperation. Creating reporting standards and agreeing to computer protocols at the cooperating institutions can be a roadblock, Brat says. Some major research centers may simply choose to go it alone, rather than turn over tissue samples and patient information to a consortium of researchers.

Moreover, as medical information technology firms flock to the field, the issue of guaranteeing the security of personal patient health remains a major concern, especially when that information is, by agreement, to be shared for research purposes.

Saltz understands the seriousness of the patient security issue. He has been personally involved in setting up security measures in the development of both caGrid and caBIG and insisting that cooperating institutions and scientists understand their responsibilities.

“We can accomplish so much more with collaboration than we can individually,” Saltz says. “And we have so much to learn from what is already available to us.”-EH


Table of Contents


Bookmark and Share

Cover


Subscribe


Subscribe for Updates
Sign up for Emory Health email