Bad 'Precision' Medicine — If Nobody Knows How It Works, Sometimes It Doesn't

The endeavor known as precision medicine, which Obama singled out in his State of the Union Address, may sound futuristic, but its been around long enough for people to have screwed it up, and badly. One of the worst medical scandals this century started with cancer researchers at Duke promising something that sounded a little too good to be true and ended with retracted papers, dashed patient hopes and lawsuits.

But precision medicine is obviously moving forward. To learn more about it, and what lessons the past has to offer, I caught up with Keith Baggerly, whose dogged investigations uncovered the problem with the Duke project. Baggerly is a professor in the Department of Bioinformatics and Computational Biology and Division of Quantitative Sciences at UT MD Anderson Cancer Center. (He is also a witness in a pending lawsuit filed by patients and their families.)

Though precision medicine has different meanings, medical researchers tend to use that term or personalized medicine to refer to the use of individual DNA differences in tailoring treatments to patients. The strategy is being driven by advances in the ability to quickly and cheaply read the sequences of code characters in DNA and by the growing use of big data to find patterns. As described in this Philadelphia Inquirer story, a number of big data cancer initiatives are gathering momentum.

The dream of precision medicine has been particularly tantalizing for cancer treatment, since cancer cells are just ordinary cells with broken DNA mutations that change the cells instructions and cause them to run amok.

And so, in 2006, cancer researchers around the word took notice when a team led by Dr. Anil Potti at Duke claimed in the prestigious journal Nature Medicine that theyd created a highly complex mathematical system that could assess a given patients tumor and determine from its genetic make-up exactly which drugs would give that patient the best odds of survival. While investigations have revealed fraud on the part of Anil Potti, many other people made mistakes in ignoring whistle blowers and allowing the technique to be used on cancer patients in a clinical trial.

While some avenues of precision medicine could lead to new, prohibitively expensive drugs used for rare subsets of patient, the Duke technique promised to chart the best course among existing treatments said Baggerly.

It would be based on the DNA in individual patients tumors. And it didnt just apply to one kind of cancer but to cancers across the board. Instead of telling a patient there was a 70% chance a drug would work to kill her tumor, he said, they could find out ahead of time if she was in the other 30% and prescribe an alternative course of treatment.

Doctors were excited and thought if the system worked, they owed it to their own patients to adopt a form of it, he said. Several groups asked Baggerly to look into it. One danger with the approach, he said, was that it was impossible to know how the technique worked. The data were so big they were measuring thousands of things per patient and there was this perception that the analysis of such data sets would be complex, he said. In most medical tests, theres some understanding of how they work. Thats true in some of the early advances in precision medicine. In some cases of melanoma, for example, theres a break in a particular gene called BRAF, and drugs that target cells with that broken gene. Theres a mechanistic understanding of how it all works.

But with the Duke project, he said, nobody has a good intuition of what 50 or 60 things are doing at once. And so there was no way for intuition to tell anyone whether it worked at all. When Baggerly started to re-analyze how the Duke researchers created the system in the first place, it didnt work. Was he using the system wrong or was there something wrong with the system?

As he investigated further, he found egregious errors that should have prevented it from working. The team had relied on cancer cell samples that had various degrees of resistance to an array of drugs. Those had been mislabeled. Some were reversed, so that the cells that were most resistant were labelled as the least.

Here is the original post:
Bad 'Precision' Medicine -- If Nobody Knows How It Works, Sometimes It Doesn't