The nature of medical research is that it is an uncertain business - sometimes the practical applications just don't work: "GDNF first showed promise as a treatment for Parkinson's patients when scientists discovered that it could boost the survival of dopamine-producing neurons - cells that degenerate in the disease - back in 1993. But so far, the results in humans have not borne out those hopes. Early trials involving injecting the protein directly into the brain showed some promise, but a second, more comprehensive trial subsequently showed no benefit. Another recent trial that used a gene therapy approach to deliver a similar compound, neurturin, showed some signs of benefit but failed in its primary goal of improving symptoms after one year. [Researchers] believe that other attempts failed because they didn't target the right tissue precisely enough. The first attempts, he said, injected the GDNF protein into the spaces near the brain regions of interest, where it failed to diffuse far enough into the brain. Infusing the treatment directly into the relevant brain tissue, he says, caused leakage into the surrounding fluid. ... They all turned out to be negative, because the delivery was never controlled ... The new trial will introduce the gene encoding GDNF into the putamen, a brain area involved in Parkinson's disease. The gene will be carried by a virus, and will be injected directly into the brain using a technique called convection-enhanced delivery, which uses positive pressure to drive fluid deep into targeted regions. The injection will include an MRI contrast agent, and the researchers will use an MRI-based imaging system to track the distribution of the treatment during delivery. ... the imaging system will allow the team to make sure the gene gets to where it's needed. ... It still remains to be seen whether GDNF really is something that helps people with Parkinson's disease."