There are signs that two new diabetes drugs are safer than

Rezulin

, according to the drugs' developers,

SmithKline Beecham

(SBH) - Get Report

,

Eli Lilly

(LLY) - Get Report

and

Takeda Chemical

of Japan.

In the clinical trials for the

Warner-Lambert

(WLA)

drug, patients taking Rezulin had a higher chance of showing elevated liver enzymes than did the control groups. In the clinical trials, 1.9% of Rezulin-treated patients (48 of 2,510) showed elevated liver enzymes, compared with 0.6% (3 of 475) who did so in the control groups. Of the 48 who showed elevated liver enzymes, 20 patients withdrew from the trials, compared with one in the placebo group. Two patients taking Rezulin exhibited signs of a drug reaction.

In the trials for

Avandia

from SmithKline and

Actos

from Lilly and Takeda, patients didn't get such elevations and there have been no such withdrawals.

In more than 5,400 patients tested using Avandia, "there's been absolutely no sign at all" of liver-enzyme elevations, says Anthony Rebuck, who heads up clinical research and development at SmithKline. "The incidence of increases are identical to placebo." In the trials, only 0.2% showed elevations that were three times the upper limits of normal, which was the same proportion as in the placebo group.

The Actos trials showed a similar incidence in both the drug group and the placebo group to the SmithKline numbers, according to Roberta Schneider, vice president for drug development at Takeda America. Actos has been tested in about 4,500 patients worldwide, 2,600 in the U.S.

No patients withdrew from any of the trials due to liver toxicity, according to company officials. Both companies say their trials did not prescreen patients or exclude ones with slightly higher liver enzymes in their blood. Both companies say their accounts of the data include all patients, whether they received the drug or the placebo or were enrolled but received neither. Both drugs cause some weight gain, which diabetics often experience when they get control of their blood sugar. Some patients retain fluid as well.

The drugs aren't just safe; they work, too, according to the companies. SmithKline tested Avandia in at least five trials it considers "pivotal," which is the term for a study that is adequate and well-controlled enough to please the FDA. In each, the drug showed a statistically significant result in lowering blood sugar both when compared to placebo and to when patients started in the trial, according to Rebuck. He wouldn't be specific about how much the drug lowered blood sugar but said, "It's very impressive efficacy." Takeda tested the drug in six pivotal trials and also had statistically significantly positive results, according to Schneider.

Both drugs are being tested alone and in combination with other diabetes drugs. Takeda tested Actos in three "monotherapy" trials and in three trials of the drug in combination with either insulin or two other types of diabetes medications, called metformin, which is sold by

Bristol-Myers Squibb

(BMY) - Get Report

and sulfonylureas, which are off-patent. Actos is a once-a-day drug; Avandia is being tested both as a once-a-day and a twice-a-day drug. A once-a-day drug is considered superior.

Even if the drugs don't cause enzyme elevation in clinical trials, it is highly likely that the panel will recommend that both drugs be used in combination with stringent liver-enzyme testing. But after eight months to a year, it is possible that the restriction will be loosened if no liver problems surface, analysts say.

Matthew Riddle, a diabetes specialist at

Oregon Health Sciences University

in Portland, Ore., who has been a paid consultant to SmithKline and Warner, says: "The FDA is under a lot of pressure to look fierce and protective here."