Sean Khozin, MD, MPH

A recent Danish study asked people to post symptoms of a medical problem on their Facebook page and ask their friends to come up with possible diagnoses. They found that the correct diagnosis was suggested in five out of the six presented cases after a median time of ten minutes. They described the responses from “relevant differential diagnoses to very silly diagnostic suggestions.”

I’m glad the researchers did not conclude that people can rely on their Facebook friends for making medical diagnoses. Instead, they suggested that people can use their Facebook friends to figure out if they should see a doctor for their symptoms. Although this is not a groundbreaking study, it points to the potential utility of crowdsourcing in medical research. The company Patients Like Me has been at this for some time, with about 125,000 patients detailing over 1000 conditions on their website.

Equi·poise, noun \ˈe-kwə-ˌpȯiz, ˈē-\

1: a state of equilibrium
2: counterbalance

Patients who enroll in clinical trials do so because they want to get experimental therapy. In a typical randomized trial with two arms, one arm of the study is either standard therapy and/or placebo and the other arm is experimental. I’m often asked by patients randomized to the standard therapy arm of a trial if they are missing out on potentially effective experimental therapy. The problem is that if we knew the answer to that question, we could not conduct the trial in the first place, given that all clinical studies exist in a state of equipoise.

For a clinical trial to continue, it must be in a state of equipoise. This means that none of the investigators, or the larger medical/scientific community, must be aware of any benefits of the experimental arm of a clinical trial over the standard/placebo arm. In other words, there must be genuine uncertainty about which therapy is better. The goal of clinical research is essentially progressive reduction of uncertainty about the effects of the experimental drug and/or increasing the level of confidence about the outcomes associated with it.

If there is a credible enough hint of benefit in the middle of a study, the trial stops and the superior therapy is offered to everyone.

Equipoise is crucial for gaining knowledge and learning new things from clinical trials. Without it, if a clear answer existed about the superiority of a particular therapy, asking patients to participate in a clinical trial that could exclude them from getting the superior therapy would not be acceptable.

And that’s how we conduct clinical research, learn new things, discover new drugs, and push the boundaries of knowledge in medicine. We live in a state of equipoise where uncertainty is a prerequisite.

This is the provocative title of a recent publication by a colleague and prominent cancer researcher at the National Cancer Institute (NCI). The main problem highlighted by this paper is whether the biologically targeted therapies that are supposed to “target” the broken cellular pathways that cause cancer worth the cost, effort, and toxicities considering the fact that many of these drugs are marginally beneficial in prolonging a cancer patient’s life.

This is not an easy question to answer because there are many factors involved in how a patient responds to treatment and we are beginning to uncover what some of these factor are. For example, the targeted cancer drug Erlotinib costs about $700,000 per QALY (quality-adjusted life years). This is a mathematical measure of the impact of a treatment. If a treatment gives a person an extra year of healthy life, that counts as one QALY.

If you give Erlotinib to patients with advanced lung cancer, about 10% will have their tumors shrink. But if you SELECT for patients whose tumors have specific mutations (called activating EGFR mutations), you can have more than 50% of patients who will have tumor shrinkage. This is the promise of “personalized medicine,” that is tailoring drug administration to fit the unique characteristics of each patient. We are now conducting a study at NCI where we first analyze the genetic profile of each patient’s lung cancer then give a drug that specifically targets that patient’s genetic abnormality.

The ultimate goal is to tailor each patient’s treatment in such a way that we can achieve a lot by giving little and avoid making too much of too little.