Scientists also are trying to use viruses that never or rarely jump into genomes, yet hang around in the cell for the long haul. Scientists at Stanford, Children's Hospital of Philadelphia and Avigen Inc., an Alameda-based biotech company, have treated hemophiliac dogs with a blood-clotting gene spliced into what may be such a viral vector.
Even if genes are inserted into the genome, there are other ways to cut down on risk.
In some trials -- such as the French case -- a patient's cells are genetically altered in a test tube, and then added back to the body. Theoretically, scientists could screen these cells before returning them, in order to weed out ones with insertions in dangerous places.
Alternatively, a so-called suicide gene could be put in the virus along with the therapeutic gene. If the cell turns rogue -- becomes cancerous, for instance -- scientists can give the patient a drug that will activate the suicide gene, and the rogue cells will then be killed. Such a strategy has proved successful in one gene therapy trial.
But the ideal solution is to perfectly fix a genetic error without introducing any extra DNA, like the deft replacement of a spent battery in a car.
One strategy coaxes the cell to repair the error using tiny pieces of genetic material as guides. Another cuts out the bad part of a gene and stitches in a good piece to replace it.
Both strategies are technically feasible but are years away from the clinic.
Ultimately, researchers say, all the animal experiments in the world won't prevent unexpected results in people.
Unpredictability is the nature of medical research. Bone marrow transplantation, chemotherapy, vaccine development -- all have hurt and killed people before being assigned to their appropriate medical niche.
"These are early days in the field, but I think it would be foolhardy to concentrate only on the dangers now," Friedmann said. "We know that patients can be helped."