Stanford University researchers have developed a potential treatment for multiple sclerosis that works on a nearly identical disease in mice.

It will be tested on humans in four to six months. If successful, it would be the first specific treatment for the progressively crippling disease that afflicts an estimated 250,000 Americans.

The treatment also would represent one of the first big medical payoffs from a burgeoning new biotechnology called monoclonal antibodies.

Multiple sclerosis, which usually first attacks people in young adulthood, begins with brief episodes of paralysis or weakness and visual problems. Over time, the episodes come more frequently and last longer until, decades later, the victim is permanently blind and bedridden.

The mouse version of M.S., which may or may not be the same as the human version, begins when certain white blood cells, called "helper T cells," depart from their usual function of attacking infectious germs. They invade the brain and spinal cord -- normally closed to them -- and attack the insulation that sheathes nerves.

Without insulation, adjacent nerves touch and, much like a bundle of stripped wires in an electronic device, short-circuit. Signals to and from the brain are misdirected or blocked.

Scientists cannot say for certain that human M.S. is the same as the mouse disease, but multiple sclerosis victims have severe damage to their nerve insulation and T cells have been found in the damaged regions, where they are normally absent.

"What we've been able to develop," said Lawrence Steinman, one of the leaders of the Stanford group, "is an antibody that attacks the T cells and keeps them from attacking the myelin sheath," the nerve insulation.

Antibodies are protein molelcules with a shape that allows them to bind only to other proteins having a complementary shape. The mechanism is analogous to that of a lock that will accept only keys of the right shape.

Monoclonal antibodies are antibodies that are identical because they were manufactured in cells that are all descendants, or clones, of a single, specially engineered, antibody-making cell.

Because the mouse disease is caused by T cells that have gone awry (nobody knows how), the Stanford group reasoned that anything that hindered T cells might stop progress of the disease. They developed monoclonal antibodies tailored to bind to a particular protein on the outer surface of T cells in the hope that this somehow would disrupt the cell's ability to penetrate the brain and spinal cord to reach myelin-sheathed nerves.

The antibodies were then tested on mice that had been injected with ground-up spinal cord tissues to stimulate their T cells into attacking the tissue as if it were a foreign invader. This procedure almost invariably induces the M.S.-like disease after about two weeks.

Steinman said that when the monoclonal antibodies were given before symptoms would normally develop, the disease was totally prevented. When given after paralysis and weakness had begun, all symptoms were reversed in 72 hours in 14 of 16 mice. In a control group receiving no antibodies, 13 of 16 mice were either dead or more severely paralyzed after the same 72 hours.

"To me these are really stunning results," Steinman said. "We can't say for sure that we'll get the same results in humans. Even if we do, this certainly wouldn't be generally available for quite a while."

Henry McFarland, a specialist on multiple sclerosis at the National Institutes of Health, cautioned that even if the mouse disease is the same as the human one, a treatment reversing symptoms in mice early in their disease still might have little or no effect on human sufferers late in their disease.

Multiple sclerosis often is not diagnosed until after a person has had several attacks over a period of years.

Steinman said the reversal of symptoms probably meant that the damaged myelin sheath had been able to repair itself, but he agreed that if the damage was too severe, it might not be possible for the sheath to recover.

The research team also includes Matthew K. Waldor, Subramanian Sriram, Richard Hardy, Leonore A. Herzenberg, Leonard A. Herzenberg and Mae Lim, all of Stanford, and Lewis Lanier of the Becton Dickinson Monoclonal Antibody Center in Mountain View, Calif. Their report is being published in the Jan. 25 issue of the journal Science.