FRANKFURT (Reuters) - Shares in Roche fell on Thursday after an early-stage obesity pill candidate that carries high market hopes was linked to a high rate of temporary side effects in its initial test phase on humans.

Roche shares were down 4% at the open at 0700 GMT after the company presented late on Wednesday details on the trial. A brief summary of the study's success in July had boosted the Swiss drugmaker's share price.

According to a presentation at a meeting of the European Association for the Study of Diabetes in Madrid, all 25 trial participants experienced mild or moderate side effects, or adverse events in industry parlance, including those that only received an ineffective placebo.

Side effects from the drug known as CT-996 - part of Roche's $2.7 billion acquisition of Carmot in December - were mostly gastrointestinal (GI), like those associated with similar drugs.

Analysts at Barclays and Jefferies described the number of adverse events as high, reflecting similar data on another experimental weight-loss drug in Roche's early pipeline.

"Investor excitement for Roche's obesity franchise may now take a pause, in our view, given both acquired assets showed higher-than-anticipated GI side effects," Jefferies analysts said.

In terms of efficacy, Roche said that the once-daily pill CT-996 resulted in a placebo-adjusted average weight loss of 6.1% within four weeks in obese patients without diabetes.

It said the Phase I trial had followed a convention of ramping up the drug dose, a process known as titration, faster than is planned for later trial stages, in order to quickly uncover any unforeseen side effects.

The frequency of adverse events was "consistent with brisk up-titration and early stage of development", Roche said.

"These data support the continued investigation of CT-996 in studies of longer duration with larger sample sizes and slower titrations," the company said.

Roche also said on Wednesday that the headline result of 6.1% weight-loss over four weeks was based on just six patients, underscoring the uncertainty of the development project.

(Reporting by Ludwig Burger, editing by Rachel More and Emelia Sithole-Matarise)