In my Oct. 13 column (with Ramesh Batchu), "Genomic Medicine Today," I noted that the pioneering Harvard geneticist George Church suggested that genomics, epigenetics, synthetic biology, stem cells and other aspects of personalized medicine would arrive sooner than people expect.


Its arrival means that the days of the lengthy and expensive clinical trial are numbered. No trial of a drug or procedure to treat a patient at the molecular and cellular levels of personalized medicine can occur without a large population of genetically identical patients, and that is impossible. In personalized medicine, the clinical trial must be replaced by direct experimentation on the person.

Two years ago, the three of us engaged in a dinner discussion about essentially this issue:

Weaver: The scientific method is very linear. There's no provision in the scientific method to leapfrog into a completely different paradigm. You can't randomize between cutting somebody's rectum out and making a permanent colostomy and doing a local incision with radiation therapy. Nobody in their right mind would ever say, "Oh, yeah, let me just subject my rectum to this random draw as to what's gonna happen."

Ellis: But we randomize clinical trial patients into those who will get a promising intervention and those who will get the traditional treatment. The researchers must have some sense — after all, it's the reciprocal of their research hypothesis — that the patients who don't get the new intervention stand a reasonable chance of poor outcomes compared with those who do.

Weaver: So how ethical is that? We've not given enough thought about giving treatment that is unproven versus withholding unproven treatment that has great promise.

Ellis: Perhaps we coddle people too much. It's almost considered unethical to allow the patient to choose, because the patient does not know enough. Why can't people have more say in their lives regardless of their ignorance? Don't they have that right?

Weaver: Absolutely they do — within an ethical hierarchy. There are the ethics that govern me as your doctor and there are the ethics of you governing yourself.

The physician has the opportunity to tell the patient: "This is what I think is likely to happen." But frankly, many times our instincts in these regards are probably little better than the patient's. And since it's the patient's own self that's at risk, it seems to me the patient has the right to choose. It is our obligation to inform them, to educate them as much as possible.

Because of the acceleration of technological innovation in medicine, we need a method that allows leapfrogging into new paradigms without being reckless on one hand or backward on the other. It's a fundamental question of what constitutes a valid scientific observation.

Ellis: We first have to realize that modern science still depends heavily on the pvalue (probability). Certainly the social sciences and to a large degree the medical sciences are based on statistics, and the problem with statistics is that they're averages based on samples representative of the study populations. But doctors deal with individuals and individuality. Of course, we're recognizing now through genomics and proteomics and all this that every patient and every disease state is very, very different.

The pvalue is becoming increasingly irrelevant. It is the first thing that we need to get away from. And we can, because we're now rapidly acquiring true population data, not merely sample data.

And with today's EMR, demographic and clinical data on every patient are being collected. So we are at the point now, or very soon will be, where we no longer need to do sampling to get representation in the population. We will have population parameters rather than sample statistics. This must surely be what the Mayo Clinic saw and began implementing a long time ago, when it digitized every single one of its patient records going back to the clinic's opening. So it's ahead of the game. Any hospital with an EMR could be in that game, by applying advanced data mining and analytic techniques to its EMR data.

Pontes: Why not look at a cohort of people who have responded to every chemotherapy that we've given and survive? How are those people genomically different from the nonresponders, and how does that affect how they interact with drugs? We tend to look at a population as if it were uniform, but more than 30 years ago Donald Coffey from Johns Hopkins knew that medicine had to be personalized! He wrote a short paper as a summation of a conference organized by the National Prostate Cancer Project in the '70s.

He described the case of a lab tech left to oversee an experiment in which he was supposed to inject hormonally sensitive cells into inbred rats from which a particular prostate cancer cell line was derived. But the young man ran out of the hormonally sensitive cells, so he simply injected the rats with hormonally resistant cells, which he had available. And then he ran out of inbred rats, so he started injecting rats from different strains.

Of course, this was a joke, but his point was that particular situation was better than any clinical trials used in humans because of the human genetic diversity in any particular clinical trial. If a therapy doesn't work for just about everybody, we reject it, when in fact we should be looking at only the cohort of people, and ultimately perhaps the individual, for whom the intervention is working.

Weaver: Is there anything written about the scientific method, the cohort studies, p values, compared with actual population numbers?

Ellis: If it has, it is not highly visible. Which is surprising, because the implications of being awash in both individual and population data are big. But nobody seems to be writing about what amounts to a paradigm shift in biomedical research or suggesting what we need to be doing today to prepare for and adopt the new paradigm.

Some hospitals are collecting lots of EMR data and could start to use it — today. There's no reason to wait. They could be delivering better care to patients today by mining that data.

Weaver: They could start by looking at all the people who responded to a treatment or all the people who lived but whom you thought should have died, or vice versa: all the people who died whom you thought should have lived.

Ellis: Leading hospitals have been capturing such data for several years already, in their EMRs. You ought to be able to push a button tomorrow and get the answers so that when one particular patient comes in, you can check that patient against your database of experience.

Weaver: Apart from its practical value, this has marketing value. If you tell people, "We will match your case against our best cases in our database so that you can get this kind of therapy" — that's appealing.

Ellis: I have read that at Mayo, the patient comes in, is seen by a team, not just an individual physician, and the first thing they do is enter the patient's data and compare him or her against their database of 100 years of patient records.

On a relevant tangent, on PatientsLikeMe (a website and database of patient-contributed clinical and demographic data) anybody can search or browse the database to see which therapies work with which patients. There are some issues surrounding it, but it's got traction. Thousands of patients have entered quite detailed information about their demographics, diagnoses and therapies.

Pontes: Researchers are under a lot of pressure to produce a lot of jargon-laden papers rather than have real-world impact. If they ignore the pressure, they are often vilified. Decades before the DNA structure was known, Barbara McClintock proposed the concept of "jumping genes" that changed the colors of corn kernels.

And they thought she was a nut! Now, her work is recognized as the basis of tumor cell heterogeneity and many other fundamental characteristics. She went on to win the 1983 Nobel Prize in physiology or medicine.

We need to go back to that ethos of genuine creativity.

Ellis: Yes, we need to recognize that the "nutcase" is worth listening to — that today's scientific method is not the be-all and end-all.

Weaver: But can a new method of scientific understanding be developed outside the box of p values and statistics?

Ellis: That is the Holy Grail. You may have a very good study and be able to say that within this level of comfort and this level of statistical significance, that research sample was prone to that disease.

Weaver: But it has no meaning with regard to an individual.

Pontes: Exactly.

Weaver: We need to be able to penetrate down to that level. How many times, Edson, does a patient ask you, "OK, I've got a certain tumor, what's my likelihood of being cured of this?"

I always start by saying, "It's impossible to say. You will either be cured or you won't be cured." I tell them, "We can take a large group of people who may or may not resemble you and can say statistically this is what is likely to happen for the average person in that group. But for you, the statistic does not answer the question. Now, it might be that by looking at your family history, your genes, your tumor, your proteins and so on, maybe we will be able to answer it."

Using advanced data mining tools, I might at least be able to tell the patient: "We had 10 people with a composition almost identical to yours, or very similar in a group that we would classify as being similar to you, and 90 percent of the people in that group were cured."

I don't know how we get to this new paradigm.

The Next Step

It seems to us now that we get to the new paradigm first by acknowledging that a new paradigm is needed. It will take what George Church called a "brave" sick person to be the guinea pig for the first personalized therapy under the new paradigm, but if the result were not just the curing of the condition but better health than the patient had before he or she became sick, personalized medicine would spread like wildfire.

Perhaps that's how we get to the new paradigm.

Before such an experiment takes place, however, the physician and the patient must be much better informed about the chance of success than they are typically informed under today's paradigm, and the chance of success must be close to 100 percent. Essentially, they will know the outcome in advance, which means there is, or should be, nothing experimental about personalized medicine at the point of care. We feel confident in asserting that personalized medicine will deliver nearly 100 percent successful outcomes because of advances in synthetic genomics and systems biology that have happened since we held our dinner discussion and that will continue to accrue.

In 2010, Craig Venter brought synthetic genomics into the limelight when he and his colleagues inserted synthetic — human-made — DNA into the shell of a bacterium, thereby creating biological artificial life. Another field, known as systems biology, also creates artificial life but in digital rather than biological form. It enables the mathematical modeling of organisms, organs and tissues from data produced by the growing number of sequenced genomes and the knowledge of how the genes work.

By modeling the condition, we seek to understand a systems biology (digital artificial life) model of the affected tissue, organ or even the whole organism. Then, digitally modeling the intervention (the drug or the procedure), we can do most of the work extremely quickly on a computer, with a strong probability that the result will be safe and effective. We can then re-create the model biologically through synthetic genomics, thereby boosting the probability closer to 100 percent.

The odds for the patient will be much better than the odds of any traditional treatment, and the ethical boot will be on the other foot: It will be unethical to practice traditional medicine when personalized medicine offers such better odds.

David Ellis is a futurist, author, consultant and publisher of Health Futures Digest, a monthly online discursive digest of news and commentary on long-range, leading-edge technological innovations and their consequences and implications for health care policy and practice. He is also a regular contributor to H&HN Daily and a member of Speakers Express.

J. Edson Pontes, M.D., is a urology professor at Wayne State University School of Medicine in Detroit, and he practices at the Karmanos Cancer Institute. He is also the head of international services for the Detroit Medical Center.

Donald W. Weaver, M.D., is the surgery department chairman at Wayne State University School of Medicine and surgeon-in-chief for the Detroit Medical Center.