A 52-year-old woman arrives at Brigham and Women’s Hospital (BWH) with headaches and a cough. After a series of tests and scans, she receives a diagnosis of metastatic lung cancer. Her prognosis is poor: her chances of living for another 18 months are less than one percent. After two rounds of traditional chemotherapy, her condition deteriorates. Researchers at BWH analyze the genomic sequence of her cancer cells and find a specific genetic mutation they can target with an existing drug.
Four years later, she has no sign of disease.
Deep in the genetic sequence of the patient’s cancer cells, scientists found information that provided clinicians with a successful plan of action. This transformation—turning massive amounts of raw data into therapeutic treatment plans—is the promise of precision medicine.
We live in an age of big data. From marketing to public health, vast data sets are analyzed to predict everything from a shopper’s potential purchases to the patterns of flu outbreaks. Big data sets are also being generated in healthcare—seemingly endless amounts of information about every patient, condition, procedure, and drug across multiple providers and organizations.
The BWH Department of Pathology, chaired by Jeffrey A. Golden, MD, performs more than seven million laboratory tests every year. The results can be as simple as routine cholesterol counts or as complex as the three billion base pairs of the human genome. Golden estimates 60 to 70 percent of all data in medical records come from laboratory tests.
Radiology is another significant source of patient data. Led by Steven Seltzer, MD, BWH’s Department of Radiology performs a million diagnostic imaging studies per year, each of which can have anywhere from 30 to 100 different images.
Precision medicine—harnessing the big data of healthcare to more precisely diagnose and treat diseases for entire populations of people—is evolving at an explosive rate, with many medical centers across the country focusing primarily on genetics and genomics. While Golden acknowledges the important role the human genome plays in precision medicine, he notes it is merely the tip of the iceberg.
Of course, we are collecting all these large datasets, all the ‘Omics’: genomics, proteomics, metabalomics, and so on. But precision medicine also includes clinical data from all over the hospital: pathology results, radiology images, EKGs, EEGs, pulmonary function tests—all the information we gather on our patients,” Golden says.
An added challenge of making meaning from big data is analyzing and integrating the vast amounts of information from a wide array of sources outside BWH. Golden notes, “We have to look at external databases, population studies, clinical trials, medical literature. We need to build an entire infrastructure, not just a big repository.”
Golden likens the required infrastructure to the ways maps have evolved. Paper maps can range in detail from showing entire countries with main highways and secondary roads to detailed street maps for a particular city. However, the Internet has enabled online maps to provide a wealth of real-time detail, such as traffic conditions, road closures and accidents, hotels, restaurants, points of interest, and even images of intersections or buildings at a given address.
These real-time data points we can now overlay onto an online map aren’t all built into the map itself. They come from other sources like traffic cameras, global positioning satellites, and mobile apps,” says Golden. “We want to provide the same kind of experience with precision medicine.”
Turning data into knowledge
Seltzer uses examples from radiology to describe the opportunities of precision medicine. “In the past, when referring physicians thought about what kinds of diagnostic imaging tests to do for patients, they relied largely on their own experience and their own clinical judgment, which was limited to the number of patients they had taken care of on their own.
When our radiology department went digital, we gained the ability to investigate data from thousands of patients who have come into our hospital with similar signs and symptoms over the past 10 years. We can determine what diagnoses they had, what imaging tests they had, and decipher which imaging tests proved to be the most effective for them.
Now, referring physicians can select diagnostic tests based on the experience of dozens of their colleagues and hundreds of thousands of patients who came before. This kind of decision support helps us to more quickly do the right thing for the right patient at the right time,” Seltzer says.
These advanced decision making capabilities in the radiology department, as well as the case of the 52-year old lung cancer patient, point to what lies ahead. BWH is on the cusp of using big data as the foundation for precision medicine, but the full potential to affect lives worldwide remains a dream without the computing power, data storage, and informatics that can transform raw clinical data into relevant information and help guide a caregiver’s treatment plan.
The task, Golden says, “is to build sophisticated algorithms that allow us to reveal the relationships all these data have with each other. These algorithms will pave the way for predictive models that will give us the ability to better anticipate disease, ultimately leading to more effective healthcare. We have started this work and expect it will proceed for the coming years. This will never be static; as we acquire new data we will be dynamically updating the ways we use them.
This is about providing caregivers more information on how best to manage their patients,” says Golden. “Information about what drugs they’re going to prescribe, what the possible complications are, when this person needs to be admitted, and when they don’t need to be admitted to a hospital.”
Building networks to discover patterns
Beyond the walls of BWH, scientific and healthcare institutions, and government agencies around the globe are generating and striving to use similarly vast amounts of medical data in comparable ways. Current efforts to mine these huge, rich data sets to extract greater insights into complex human diseases—like cancer, heart disease, diabetes, neurodegenerative disorders, and many more—are in their infancy.
One piece of this big data puzzle, genome sequencing, has not produced a paradigm shift in medicine for one simple reason: while our genome provides clues to what can happen to our health, it is not a reliable indicator of what actually does or will happen. Two people with the same genetic mutation may have very different health profiles.
The classification strategy we use for human diseases goes back to the 19th century,” says Joseph Loscalzo, MD, PhD, chair of BWH’s Department of Medicine. “Historically, diseases have been defined, treated, and prevented as inclusively and broadly as possible. For example, anyone with blood pressure above a certain level is considered to have hypertension. But hypertension is a syndrome that comprises many different diseases of many different causes that present differently in different people.”
We’ve also learned a lot by studying well-known, classic genetic diseases, such as sickle cell anemia,” Loscalzo says. “It’s a disease caused by a single mutation, but there are several different clinical syndromes with which a patient who harbors the mutation can present. This range of clinical presentations reflects the genetic context within which the sickle hemoglobin mutation exists, with other adaptive genetic variants offsetting the effect of the primary mutation and other variants worsening the effect. It is far too simple to claim that one mutation leads to one clinical disease. The biology of human disease is far more complex.”
To gain further understanding, Loscalzo says, “We need to distinguish subgroups of diseases by their molecular signatures. Unfortunately, the molecular signatures tell us nothing about how one gene or gene product interacts with another. This is where network medicine comes in. Network medicine describes how the different molecules in the body interact with each other, the way an assembly diagram explains how car parts fit together to become a car. You wouldn’t build a car without an assembly diagram. You wouldn’t know what to do with all the parts.”
By drilling down and understanding this network, or interacting patchwork of molecules that are the origins of disease, clinicians gain a better ability to give patients more precise diagnoses and develop targeted treatments. The network, Loscalzo adds, is a metaphor for the questions precision medicine aims to answer, as well as the methods by which investigators are working toward solutions. “To understand the networks underlying disease, we need networks of professionals with complementary skills to address these questions,” says Loscalzo.
Golden has assembled a team of experts in cardiology, pathology, radiology, oncology, pulmonology, epidemiology, gastroenterology, and genetics to form a Precision Medicine Committee. With representatives from BWH and other elite institutions such as Dana-Farber Cancer Institute, The Broad Institute of Harvard and MIT, and the Countway Library at Harvard Medical School, the committee is developing its first precision medicine initiatives in three areas: cancer, pulmonary fibrosis, and inflammatory bowel disease.
Eventually, Golden envisions an Institute for Precision Medicine at BWH, a first-of-its-kind platform that will realize the potential of leveraging massive data sets to transform healthcare delivery around the world. The committee is currently laying the ground work for the new institute, with the goal of launching the first two clinical programs within the next year. Such an institute would unite some of the world’s most gifted medical minds with systems biologists, experts in bioinformatics, applied mathematicians, biostatisticians, physicists, and computer scientists to decipher the complexities inherent in precision medicine.
While aspects of this work are already taking place in pockets throughout BWH, an institute would unify these efforts and provide faculty with the resources—such as biostatistical and bioinformatics support—necessary to help translate reams of data into useful clinical information to benefit patients with a variety of health issues.
From snapshots to trends
Today we are missing a huge opportunity to provide better care for our patients, even with the data we already collect,” says Golden. “The amount of information we have on our patients is enormous, but the way it’s currently used is that one test is ordered and that single result is provided back to that clinician.”
Rather than using test results to get a snapshot of a patient’s current health profile, Golden believes precision medicine will be able to provide caregivers with insights on how a patient is trending.
Right now, it has been shown that if you give a group of expert doctors the result of just one test—a hematology panels—for 10,000 patients, they will only be able to tell you if that patient is anemic or not,” says Golden. “However, if you build the right algorithm, as has been done, a doctor can be provided, with great accuracy, who is going to be anemic in two months.”
Instead of a snapshot, precision medicine will give clinicians a moving, predictive picture of a patient’s health.
You can know who to treat and, more importantly, who not to treat. This will prevent patients from getting a drug, even something as simple as iron, which has definite side effects, because we know they’re going to be okay in two months. At the same time we treat those patients that we know are likely to get in trouble, and thus prevent them from becoming anemic,” says Golden. “If you keep patients from becoming anemic, you prevent the potential complications associated with this condition, like heart attacks and strokes; and you keep them out of the hospital, eliminating the chance of them getting hospital acquired diseases like secondary infections. Most importantly, you keep individuals productive, you keep them at work. This is what we can do now with data we’re already collecting.”
A new healthcare paradigm
Precision medicine will provide caregivers with previously unimaginable microscopic and macroscopic pictures of their patients. It will fundamentally change the way they provide care and develop prevention strategies, for individuals and entire populations of patients, ultimately reducing unnecessary hospitalizations and keeping society healthier and reversing the trend of skyrocketing healthcare costs.
Precision medicine will be like the iPhone,”says Golden. “When the iPhone first came out in 2007, there were already cell phones, electronic calendars, portable music players, and electronic devices that could surf the Web. The iPhone was a disrupter because it put all those capabilities in one device. Precision medicine will be a similar disrupter
for healthcare. It will change the paradigm by
improving the health of individuals and populations,
providing better care at lower costs.”