* Currently no approved medicines to slow Alzheimer's
* Crenezumab fails to meet main goals in mid-stage study
* Signs of benefit at higher dose in milder patients (Adds comment from analyst, independent researcher, shares)
By Caroline Copley
ZURICH, July 16 (Reuters) - Roche's experimental drug crenezumab failed to delay a decline in thinking and memory skills in people with Alzheimer's disease, a result likely to bolster a growing belief that drugs need to be given in earlier stages of the disease to show a benefit.
The Swiss drugmaker's treatment was tested in patients with mild-to-moderate forms of Alzheimer's, a fatal, brain-wasting condition that gradually robs patients of their ability to think and care for themselves.
Results of a Phase II study involving 431 patients found crenezumab failed to significantly slow cognitive and functional decline compared to placebo, missing the study's two main goals, Roche said in a statement on Wednesday.
But an exploratory analysis of patients with a milder form of the disease who received a higher dose of crenezumab via an intravenous infusion showed a statistically significant reduction in cognitive decline, Roche said.
Carole Ho, director of Early Clinical Development at Roche's biotech unit Genentech, told Reuters she was encouraged by the data, even though it missed its main goals, since it demonstrated that treating the disease earlier could increase the benefit.
Ho said Roche would decide on any future plans for additional clinical studies following an analysis of the data in conjunction with health authorities.
Analysts had anticipated limited success for crenezumab, after a similar treatment from Pfizer Inc and Johnson & Johnson called bapineuzumab, and solanezumab, a drug from Eli Lilly and Co, failed in late-stage trials.
All three drugs work by blocking the toxic protein beta-amyloid that forms plaques in the brain believed to signal the onset of the disease.
Another protein called tau that forms twisted fibres and tangles inside the brain is also considered to be a possible culprit.
Sanford C. Bernstein analyst Tim Anderson said the positive impact in patients with mild Alzheimer's lent credence to the beta-amyloid hypothesis but added this would need to be vetted further in large Phase III trials.
But some researchers were sceptical about pursuing the amyloid theory.
"This failure combined with a dozen or so prior failures are definitely a blow to the amyloid hypothesis as we originally proposed it," said P. Murali Doraiswamy, a professor at the Duke Institute for Brain Sciences. "The hypothesis needs to be changed and we need to pursue non-amyloid targets, or else we will suffer more failures."
Based on the results of the trial, the odds of crenezumab gaining approval for treating mild to moderate Alzheimer's are very low, he added.
Shares in Roche closed down 0.6 percent at 266.40 Swiss francs, underperforming a slightly firmer European healthcare sector index
The findings add to a string of data suggesting the best hope is testing drugs much earlier in the process before patients' brains are wrecked by Alzheimer's.
Lilly has since started a new clinical trial focusing only on patients with mild signs of the disease.
Other trials are already underway testing people who have not yet shown any symptoms of Alzheimer's to try and gauge whether early intervention can prevent or slow the disease.
Crenezumab, which was licensed from Swiss biotech company AC Immune in 2006, is also being tested in a U.S. government-backed trial in a group of Colombians with a genetic mutation that causes them to develop Alzheimer's early. Results of that trial are due in 2020.
A second Alzheimer's drug from Roche, known as gantenerumab, is also being investigated in a late-stage trial with patients who are yet to develop any signs of the disease.
Roche said a smaller Phase II biomarker study also showed an effect of slowing cognitive decline in patients with a milder form of the disease. Details of this study will be presented at the Clinical Trials in Alzheimer's Disease meeting in November.
Jeffrey Cummings, director of the Cleveland Clinic Lou Ruvo Center for Brain Health, who presented the crenezumab data at the Alzheimer's Association International Conference in Copenhagen, said he was encouraged by the consistent benefit seen in the most mild patients.
"It has been very difficult to have any convincing evidence of a treatment effect in Alzheimer's Disease," Cummings said. "When you see this kind of consistency across two trials then you would definitely want to advance a drug in a trial that focuses specifically on very mildly affected patients."
The drug was well tolerated with only one case of vasogenic edema, a brain swelling side effect seen in similar drugs, allowing crenezumab to be administered at higher doses, Roche said, though patients taking the medicine did have a higher incidence of pneumonia than those on placebo.
At lower doses when crenezumab was administered through an injection beneath the skin no significant benefit was seen even in milder patients, the results revealed.
Alzheimer's - the most common form of dementia - already afflicts 44 million people worldwide and this figure is set to triple by 2050, according to campaign group Alzheimer's Disease International.
Unlike heart disease and cancer, which have seen major strides in drug development, no new therapies have been approved to treat Alzheimer's in a decade, according to a recent study by researchers at the Cleveland Clinic. Current drugs only treat symptoms despite years of research.
A startling 99.6 percent of clinical trials in Alzheimer's failed between 2002 and 2012, the Cleveland Clinic study found. (Reporting by Caroline Copley; Editing by Elaine Hardcastle and Jonathan Oatis)