The disease process does not attack the macula's light-sensing cells directly, most experts concur. It starts in the layer of tissue that lies just below them. The cells that constitute this tissue are known as RPE, or retinal pigment epithelium, cells. Like worker bees tending a hive, these cells provide the light-sensing cells with nourishment and dispose of their wastes. But in contrast to many other types of cells--skin cells, say--adult RPE cells cannot replace themselves through cell division. Thus when the RPE cells begin to sicken and die, so do the cells they support.

About five years ago, leading ophthalmologists began exploring the possibility of replacing the dysfunctional RPE cells with healthy fetal cells. In theory, says Columbia University ophthalmologist Dr. Peter Gouras, "it makes a lot of sense." The RPE cells form a single layer, rather like tiles on a bathroom floor, he observes. "So why not just go in and repave that layer with new tiles?" Fetal RPE cells seem ideal for the purpose. Unlike their adult counterparts, fetal RPE cells can divide and thus increase in number. Also, they are likely to continue functioning for a number of years. And, importantly, because they are immature, fetal cells should provoke little or no response from a transplant recipient's immune system, thus making rejection less likely. Or so experts reasoned.

But how would fetal-cell transplants work out in practice? In 1993--immediately after U.S. President Bill Clinton lifted the ban on fetal-tissue research imposed by the Bush Administration--Ernest, for one, launched a project designed to find out. In the beginning, he remembers, the technical challenges seemed overwhelming. He and his colleagues were not even sure whether fetal RPE cells could be kept alive in laboratory cultures long enough to make transplantation feasible. Then, after a young physician demonstrated that this could be done, the research team began a series of experiments to determine the best way of delivering fetal RPE cells to patients.

Early on, the researchers rejected the simplest method--suspending the cells in solution and injecting them into the eye--because cells handled in this fashion did not grow particularly well. The team found that it obtained much better results when it attached the cells to a sticky substrate like fibrinogen, a protein involved in blood clotting. "And then," says Ernest, "we made a serendipitous discovery." Dr. Karine Gabrielian, a physician on the team, had been struggling to fashion the thinnest possible slivers of fibrinogen. Checking on her samples one morning, she found that some of the slivers had curled up into spheres, each the size of a coarsely ground speck of pepper. Gabrielian added several of these odd-looking constructs to a culture dish that also contained fetal RPE cells. Within 24 hours, the cells attached themselves to these motes of material and started to grow. Then the researchers transplanted the spheres into the eyes of rabbits, positioning them just beneath the retina. The RPE cells did not stay put; instead they migrated throughout the eye. This suggested that it should be possible to position a transplant at a safe distance from the macula and still get therapeutic results.

But as his team made progress on one front, Ernest grew increasingly worried about the immune system's response to the transplants. Contrary to what many had supposed, fetal RPE cells did not behave as if they were immunologically neutral. In experiments in Sweden, for example, transplanted cells were rejected. And Ernest's team found that adding fetal RPE cells to laboratory cultures sent white blood cells, which attack transplanted tissue, into overdrive. Curiously, however, adding even greater numbers of RPE cells to the culture appeared to force the white blood cells into a quiescent state, thus lowering the chances of rejection. Pearl Van Vliet's transplant, accordingly, contained a souped-up 250,000 fetal RPE cells.