There's an unusual bit of good news out of the U.S. Food and Drug Administration. In March 2010 and again last November, the FDA rejected approval for a new medical device designed to help doctors detect cancerous skin lesions caused by melanoma, even though the agency's scientific advisory board voted in favor of approval both times. But yesterday, the manufacturer, Mela Sciences, announced that the FDA had reversed course and agreed that the device is in fact approvable -- pending agreement on labeling and the compilation of an appropriate user manual for physicians.
That's good news. About 70,000 people are diagnosed with melanoma in the U.S. every year, and nearly 9,000 die from it. But it is nearly 100 percent curable if detected early. The question remains, however, what caused FDA to change course?
Unfortunately, it's difficult for doctors to distinguish between skin lesions that are cancerous and those that are benign, so the detection process usually involves doctors making a judgement call about which lesions to biopsy based on their appearance. Highly experienced dermatologists are fairly accurate, but less experienced ones and general practitioners aren't nearly as good. That, unfortunately, leads to a lot of false positives (unnecessary biopsies that are expensive and at times painful and disfiguring) and false negatives (lesions not biopsied that are actually cancerous). The MelaFind device aids in detection by digitally scanning a lesion using light from multiple wavelengths, including infrared, and then analyzing its appearance with an algorithm that factors in all the relevant characteristics. As an output, it essentially provides the doctor with a second opinion about whether any individual lesion should be biopsied. It can't replace a doctor's educated judgment, but it does provide a pretty good backstop against possible mistakes.
The FDA had cleared the clinical trial design and stipulated in advance that it would approve the device if it met certain goals. Nevertheless, the agency changed its mind after the results were in and rejected approval on the grounds that the device did no better than experienced dermatologists in detecting cancerous lesions. Like the dermatologists in the study, MelaFind did not detect 100 percent of the melanomas. And the study didn't measure how well a typical (i.e. non-experienced) dermatologist would do while using the device. Essentially, the FDA was letting the perfect be the enemy of the good. The MelaFind was 98.4 percent effective in identifying melanomas, whereas typical dermatologists in a comparison arm of the study had just a 78 percent success rate. But if it couldn't outperform the very best doctors, the agency seemed to conclude, it wouldn't let anyone use the device.
At this point, a typical medical device or pharmaceutical company would have gone back to the drawing board or have given up the product altogether. Fortunately, Mela Sciences didn't fold so easily. After the FDA's initial rejection last March, the manufacturer filed a citizens petition requesting a reevaluation. And they also went public, charging the agency with changing the rules in the middle of the game. That drew a lot of attention, and the MelaFind controversy was featured in a House Energy and Commerce Committee hearing this summer.
Fortunately, those efforts payed off. FDA won't say why it changed its mind, but I don't doubt that congressional oversight pressure made this kind of behavior untenable. Now, if only Congress would see this as an example of FDA's systematically flawed decision-making and change the rules of the game, then we'd all be better off.