By Mark Wolverton
Photos by Addison Geary
For most of medical history, practitioners have faced a great dilemma. While in most ways, people are pretty much alike, they can still be very different on an individual and personal basis: not merely in superficial things such as skin color, hair color, height, weight, or even personality, but on a more profound genetic level. For physicians, that means that while the procedure for treating something like a broken arm is basically the same for every patient, dealing with a disease such as cancer is decidedly not. One is simple and straightforward; the other is infinitely more complex. Yet until very recently, given the tools and techniques available to doctors, a one-size-fits-all approach – aimed at the supposedly “average” patient, possibly with only minor, imperfectly understood variations – was the only option.
The Human Genome Project, the multibillion-dollar effort to map the entire human genetic blueprint, changed all that. The wealth of information now available to doctors and researchers, along with new molecular biology techniques of unprecedented power, promises to supplement and perhaps even replace the coarse instruments of traditional medicine with new tools that can be wielded with exquisite precision and sensitivity.
The new term is precision medicine, one that few in the general population had ever heard until President Obama’s 2015 State of the Union address. During his address, he called for a national Precision Medicine Initiative. “What if figuring out the right dose of medicine was as simple as taking our temperature?” he asked. It may not be quite that simple yet, but Penn researchers are at the forefront of the effort to realize the promise of this extremely new and still largely unexplored approach to medicine.
Even if the strategies and techniques now being developed are new, the idea itself really isn’t. Terms such as “personalized” or “individualized” medicine have been around for a while, representing the notion of approaching and treating each patient as an individual with a unique set of problems and issues. In that sense, of course, all medicine has more or less been “personalized” from its very beginnings: As Obama observed, “Doctors have always recognized that every patient is unique.” But there’s a big difference between that and what’s now called precision medicine – a difference that’s, well, very precise.
“It’s a way to encompass personalized medicine without calling into question the fact that medicine has been personalized for a few thousand years,” says David B. Roth, M.D., Ph.D., chair of the Department of Pathology and Laboratory Medicine and director of Penn’s new Center for Precision Medicine (CPM). “It has to do with technological advances and being able to measure things very precisely and also to be able to make very highly targeted approaches to diseases like certain cancers, where we haven’t been able to do that before.”
The main reason we have not been able to do such things before is, in a word, information. Our ability to store, use, and disseminate it was limited. When the Human Genome Project completed its work, some observers – many of a more entrepreneurial than scientific outlook, hailed the achievement as the dawn of a new age that would rapidly lead to definitive cures for cancer, heart disease, AIDS, and nearly every other scourge of humanity. After all, now armed with this new deep insight into the essentials of life, how could we not solve nearly every medical mystery?
David B. Roth, M.D., Ph.D., director of the new Penn Center for Precision Medicine, meets with Guannan Wang, Ph.D., a postdoctoral researcher in molecular and cellular biology.
Nature, as usual, turned out to be far more complex and stubborn than we’d hoped. Genes are indeed important, perhaps of primary importance in human health and some diseases, but it’s become increasingly clear that curing cancer, for example, is not simply a matter of turning the right genes on and off. A crucial insight of precision medicine is that more than just genetic information must be considered. It’s one reason that the earlier notions of “personalized medicine” using a person’s genome as the sole guide to curing his or her disease have evolved into far more sophisticated and broad-based concepts. The new medicine offers the prospect not just of treating existing disease but avoiding it altogether by identifying and flagging a person’s individual risk factors.
“Humans are not hardwired by their genomes,” noted an editorial in Science Translational Medicine last summer. That’s why, said the authors, “a defining assertion of precision medicine is that genomics – no matter how powerful or economical – is far from sufficient to understand human physiology and pathophysiology. Myriad other components – molecular, developmental, physiological, social, and environmental – also must be monitored, aligned, and integrated in order to arrive
at a meaningfully precise and actionable understanding of disease mechanisms and of an individual’s state of health and disease.”
But even if the initial hype over the sequencing of the human genome turned out to be overly optimistic, it has led to definite new advances that are now driving precision medicine. Of particular note is the development of next-generation, high-throughput sequencing technology that has sharply reduced the cost and increased the efficiency of genomic analysis. For that reason, the concepts of precision medicine are concentrated at present on a problem where they can do the most immediate good: fighting cancer.
As Francis Collins, M.D., Ph.D., director of the National Institutes of Health, and Harold Varmus, M.D., wrote recently in The New England Journal of Medicine, “Oncology is the clear choice for enhancing the near-term impact of precision medicine” (2/26/15). It’s an arena in which the more finessed approach of precision medicine can be a distinct advantage over the blunt force of more conventional treatment strategies.
“The way we traditionally have treated cancer is carpet bombing,” says Roth, using a self-consciously militaristic but apt metaphor. “There’s a lot of collateral damage. You’re basically napalming the jungle so you can get the bad guys. And that’s why with traditional chemotherapy, your hair falls out, you throw up, all the classic bad side effects.” But, he adds, “if you can find those little broken bits in the [genetic] machinery and target them, you’ll generally get a medication that doesn’t cause a whole lot of side effects.”
Dr. Elenitoba-Johnson inspects a flow cell, which typically contains 30 to 40 DNA samples at a time.
Fighting Cancer, Still a Rapidly Moving Target
Compared to more systemically based disorders such as hypertension, “cancer is a relatively better understood paradigm to study,” says Kojo Elenitoba-Johnson, M.D., the Peter C. Nowell, M.D., Professor in the Department of Pathology and Laboratory Medicine and inaugural director of Penn’s Center for Personalized Diagnostics (CPD). Defining the genetic makeup of different cancers is a primary mission of the center. “We aim to deploy large-scale genetics testing in informing patient care,” he explains. Since the CPD opened its doors in 2013, more than 4,000 tumor samples have been sequenced. Because the specific genetic mutations driving certain types of cancers are now known, “there are directed therapies that can be deployed to arrest the growth of the cancer cells in a targeted and specific way without harming the normal cells.” And because genetic mutations are shared across a multitude of cancers, a therapy that targeted a mutation in one form of cancer may work against another form.
While cancer, genetically speaking, may be a relatively low-hanging fruit, it’s also a rapidly moving target. One patient’s lung cancer is not necessarily exactly the same as another’s, even if the same mutations initially sparked both tumors. Even in the same patient, cancer can mutate to evade the onslaught of targeted chemotherapy medications. That means the recurring tumor must be repeatedly sequenced. “It’s possible that in the recurrence there might be acquisition of mutations that are targetable,” says Elenitoba-Johnson. “Actually, we have examples of that, in which patients have responded famously well to the new medicine that was specified by the genetic testing.”
The center’s work, however, is only beginning. According to Elenitoba-Johnson, “The rate at which we’re able to identify mutations unfortunately is faster than the rate at which the drugs can be developed for specific mutations.” Still, he notes, “this is a new day. In the past, the armamentarium of drugs that could antagonize different mutations was a lot smaller than it is today. We have more tools in the toolkit than we’ve had in the past, but clearly there remains a lot to be done.”
Dealing with data is perhaps the main obstacle to the promise of precision medicine. More than simply collecting the vast amounts of information, the key to making it useful is understanding how one data point relates to others, and what it all means for treating disease.
Dr. Elenitoba-Johnson confers with David B. Lieberman, M.S., L.C.G.C, technical manager for the Center for Personalized Diagnostics.
The “Engine” Behind Precision Medicine
Creating the CPD was an element of the Perelman School’s faculty-led “Shaping the Future of Medicine: Five-Year Strategic Plan, 2013-2017.” Another recommendation was establishing the Institute for Biomedical Informatics, which also opened in 2013. “Informatics permeates every aspect of precision medicine,” says Jason H. Moore, Ph.D., the institute’s director. “Every aspect of precision medicine benefits from informatics, from the clinical informatics side, the clinical databases, the patient resources that we can tap into to identify patients at high risk, and computational methodologies to extract the data, work with the data, and find the patterns. That requires informatics infrastructure to track the samples, make sure the samples are linked to the patient, and mine the patient data.” When the researchers have identified a subgroup of patients, they are then able to identify their samples and do specific testing on those samples.
It’s far more than simply genomics and knowing the roughly 25,000 protein-encoding genes of a human being. Again, it comes down to information, whether contained in the genome or the countless other environmental and experiential factors that shape the individual human being throughout a lifetime. For precision medicine to truly achieve its full potential, that information needs equal consideration with the genomic data. Ironically, measuring and understanding the unseen subtleties of the genome is currently far easier than understanding the effects of environment. Such effects involve
not only air and water quality and toxic exposures, but the smaller-scale effects of an individual’s diet, lifestyle, activities, even psychology.
The medications a person takes can also be considered part of their individual “environment,” Moore explains. “You can think of drugs as environmental agents as well, because when you give a drug, you’re putting a chemical in the body, and there’s a whole cellular physiologic infrastructure for dealing with that. Each of us differs in our ability to process and metabolize drugs and in our response to those drugs. So we have to do the research to figure out how those multiple factors interact with each other to impact health and drug response.”
The environmental questions, Moore argues, need much more attention. “The tools and the technology just aren’t there yet. We’ve invested a lot of money in measuring the human genome and we can do that pretty well now, but we need a similar level of investment and commitment to measuring the environment and then being able to integrate all of that data using informatics into the genomic and clinical data.” In fact, Moore says, it’s not even that the research hasn’t been done yet. “We’re missing key pieces of information and data that we need to complete that research.”
One promising solution for collecting such data that’s just now becoming widespread consists of wearable or implantable personalized devices that can provide 24-hour real-time data on an individual’s various physical parameters. Many people already own devices like Fitbits, from which data could be collected and integrated into personal medical care. Specialized devices can also be designed and customized for specific studies. At Penn, Moore points out, “the Center for Excellence in Environmental Toxicology is talking about strapping on monitors that you wear throughout the day, just like an Apple watch, that measures what you’re exposed to and grabs all that data and feeds it into a database.” Such concepts are still in the very early stages, but they can be a practical means of gathering data for precision medicine.
In general, as far as patients are concerned, precision medicine is basically a behind-the-scenes matter. The CPM and CPD don’t treat patients directly: “Seeing patients and dealing with whatever advances we can help to enable will be done by the people who have always done it,” Roth says. Elenitoba-Johnson adds, “What we do is provide the clinicians and oncologists with information that basically specifies what kind of treatment their patients should get.” But there is a notable exception that makes precision medicine about as hands-on as it can get.
Sunil Singhal, M.D., top right, leads a surgery with a new tool: fluorescent dyes that make cancer glow with color under infrared light.
Knowing Better Where to Cut for Cancer
Cancer patients may have their tumors genetically sequenced and closely imaged with CT, MRI, and PET scanning. But when the time comes for a surgeon to actually remove the tumor from the body, the only tools he or she has available are the eyes and fingers. Cancer, however, is endlessly insidious and can hide in places that can’t be seen by the naked eye or palpated with the fingers.
“For the last 200 years, all you had was your hands and your eyes to decide where to cut,” says Sunil Singhal, M.D., an assistant professor of surgery. The director of the Penn Medicine Thoracic Surgery Research Laboratory, he was appointed director of the recently established Center for Precision Surgery. But surgeons now have a new tool: fluorescent dyes that make cancer glow with color under infrared light.
“What happens is a patient comes in a hour or two before surgery, and we just inject a dye into their veins, and then you give it some time for the dye to go around their body. And then once it gets taken up by the tumor, you go for surgery, and once the light in the room hits the tumor, it starts glowing.” That allows not only the margins of the tumor mass to be sharply delineated but can also reveal the presence of any satellite lesions in surrounding healthy tissues that might elude the naked eye. “By having the dye, the tumor glowing, you can know better where to cut. We’ve been able to find smaller and smaller lesions that typically would have been missed.”
Over the past three years, Singhal and his colleagues have performed more than 300 surgeries on various types of cancer using the technique, known as intraoperative molecular imaging. “There are different surgeons doing different specialties. There are some unifying themes, like finding tumors and making sure the margins are good. Each specialty has its own challenges. For example, breast cancer has a hard time with margins, and something called vector carcinoma in situ, VCIS. So we use it for that. With kidney cancer, we know where the exact margins of the kidney are, and instead of taking out the whole kidney, maybe we can just take out a chunk of the kidney, so kidney functions are not as affected. The future of cancer surgery is really transforming in front of our very eyes.” The Center for Precision Surgery is also embarking on clinical trials and research to investigate techniques and establish the efficacy of precision surgery in different varieties of cancer.
The example of Singhal’s precision surgical techniques both demonstrates the breadth of disciplines encompassed by the “precision medicine” term and highlights why Penn is extremely well positioned to make great strides in the field. “I think Penn Medicine is ripe for this kind of work,” Roth says. “We have the medical school and all the other schools such as Engineering, and we have strong computational people on campus. All the ingredients are at hand, but the scientists and clinicians are too busy doing their day jobs to really be able to make this happen.” On becoming director of the Center for Precision Medicine, Roth decided to bring all the pieces together. “I thought, let’s just add a little bit of extra power so they have the bandwidth to do it.”
With the various centers up and running, Roth believes that all of the pieces are now in place. “I’m hoping it’ll be especially useful for the research scientists, who are coming up with all these great ideas and technologies and systems but don’t have time to go to the clinical meetings where we talk about patient problems, so they’re often not aware of specific clinical problems that could be solved by their side. Even if they are aware that there’s a clinical need and have a great idea how to meet it, they don’t generally have the clinical context to even know where to start about making it happen in actual patient care. So our policy will be to provide people who know how to do that.”
That policy can lead to surprising partnerships, Roth points out. “The first project that we’ve had success with is to look at a very rare disease in humans, a very rare but lethal disease called angiosarcoma, cancer of the blood vessels. In collaboration with folks at the School of Veterinary Medicine, we realized that large breeds of dogs like German Shepherds and Golden Retrievers have these angiosarcomas very commonly.” After doing genetic profiles on the dog angiosarcomas, they identified some potential candidates as molecular drivers of the cancer, which might lead to some clinical trials on dogs in the very near future. That could be good for the dogs, Roth continues, “but more broadly, we already know from our initial work that the human angiosarcomas and the dog angiosarcomas definitely have some similarity at the molecular level. If you identify an FDA-approved drug that has significant effect in the dog angiosarcoma setting and find that driving mutation in a human, you could potentially accelerate the rate of doing some trials in those people.”
Jason Moore, Ph.D., director of the Institute for Biomedical Informatics, left, and Paul M. Kopec, M.S., its project director, sketch out ways to make “big data” useful to researchers and clinicians.
Some Hurdles Ahead
Still, there are definite hurdles to overcome before “precision medicine” becomes simply “medicine.” One is a problem of public perception that crops up with nearly every other major medical advance: the hazards of making big promises too soon. “There is a lot of buzz about [precision medicine] in the commercial space,” Roth says. “You know that not all that stuff’s going to last. I do think that there’s a danger of overhyping and in disappointing people.”
Roth believes that the answer is to “get practical” by moving precision medicine techniques from the lab into the clinical setting. “Many of our peer institutions are focusing on the research side, discovering more molecular drivers of this or that disease so we can find drugs or make better predictions about what’s going happen with an individual patient.” Penn Medicine is aiming in a different direction. “I thought, let’s come up with some wins and show that we can do this in the context of regular health care, not just some exotic trial where we need to control all the conditions very carefully. We can show some practical ways that we’re really changing the way we do this with fairly large numbers of people.”
Instead of being a “magic bullet” that will cure all human disease overnight, precision medicine is poised to develop and evolve into the general practice of health care. As Elenitoba-Johnson says, “it’s not unlikely that the whole concept of precision medicine itself will get refined as the individualized aspect becomes more prominent. As long as we are evaluating the patient within the context of the disease and their circumstances and treatment is administered in that regard, we’ll be making the right decision. For every disease, we can come up with uniform protocols within discrete population cohorts that will be appropriate to treat those patients – and then discern the appropriate treatment.”
Viewing the question from his informatics perspective, Jason Moore is somewhat more cautious. “We’re in the early days. There’s a lot of research to do, a lot of infrastructure development. It’s a fundamental change in the practice of medicine, and it will require some time to sort out. I would say we have at least another decade, maybe two decades, of research to really get a handle on what the precision medicine strategies are that we should be using.”
Jason Moore and Jing Li, a postdoctoral researcher, are interested in visualization methods that can greatly enhance the ability to make sense of data-mining results.
In a recent “Sounding Board” piece in The New England Journal of Medicine, in fact, J. Larry Jameson, M.D., Ph.D., dean of the Perelman School, and Dan Longo, M.D., suggested another impact of precision medicine: “Medical school curricula will need to focus even more on information management” (5/27/15).
According to Gail Morrison, M.D., G.M.E. ’76, senior vice dean for education, precision medicine will require that doctors classify individuals not by diseases but by subpopulations. The subpopulations will be grouped by their susceptibility to a particular disease; by the biology and/or prognosis of the disease they develop; and by the individuals’ responses to a specific treatment. That is why molecular profiling for patients in routine clinical settings will be necessary for matching patients to their ideal treatment.
To prepare the medical students, the Perelman School has implemented a required curriculum block for all first-year students in cancer biology. The new block integrates concepts important to understanding and practicing precision medicine by doing molecular genetic profiling for cancer, evidence-based medicine for making decisions, and epidemiology of various types of cancer. “We want to assure that our students will be prepared to treat patients with precision medicine as it becomes the norm in the clinical setting,” Morrison says.
Another issue that needs to be resolved is not scientific, but financial. Who pays for the tests and treatments based on precision medicine? No matter how advanced and sophisticated the techniques may be, they’re essentially irrelevant if they remain inaccessible and unaffordable to the patients who need them. “We see that with a lot of things,” says Elenitoba-Johnson. “The fee schedules are just being worked out on a national level. Sometimes local negotiations with the payers also inform the way in which the testing gets reimbursed. So that’s in flux, and the extent or what proportion of the testing you can get back is also in flux.”
As Moore puts it, “At least in the short term, this is going to be a technology and a methodology that is available only at the bigger, more well-funded institutions that can invest money on the front end. When we drop the ‘precision’ from ‘precision medicine,’ I think it’ll be a commodity that anybody can take advantage of just like any other medical technology.” On a practical level for the patient, however, it depends on whether the insurance company will reimburse for the environmental testing, the genetic testing, and the other clinical measures necessary to assemble the patient’s information. Then that information would have to be fed into a computer algorithm, which would, Moore says, “predict what drug they should get or what prevention strategy you should implement. When insurance companies start reimbursing for these tests, then I think it’ll become commonplace.”
Costs can also be expected to come down as precision medicine continues to become ever more precise. According to Roth, “If you can really precisely identify who needs the intervention, you don’t have to give it to everybody, and I think that’s going to be a great way to offset costs as we move away from volume-based to value-based payment for delivering medicine. Figuring out what’s wrong at the molecular level and having a great precise tool to treat it should lead to cost savings in a number of areas.”
However long it may take for precision medicine to shake free of its adjectival modifier, Roth and his colleagues are confident that it’s the new paradigm for the future of medicine. “I think precision medicine is the next iteration in the evolution of health care,” he says. Despite its current heyday as “the next big thing,” it seems clear that it’s not going away. “I promise you ‘the next big thing’ is going to come along, but I think we’ll make some really important advances with what we’re doing now."