News Focus
News Focus
Replies to #36513 on Biotech Values
icon url

DewDiligence

10/30/06 9:43 AM

#36515 RE: drbio45 #36513

EMIS – The highest of three doses tested showed a benefit vs placebo, but I question whether the benefit deserves to be called statsig given that multiple doses were tested and the p-value of the best arm was 0.037. I plan to listen to the CC for more details.
icon url

drbio45

10/30/06 9:59 AM

#36518 RE: drbio45 #36513

There were 35 patients enrolled in each oral insulin arm of the trial, and 36 in placebo. Results showed that the total number of patients achieving a hemoglobin A1c (HbA1c) decrease of more than 1.1% among the 35 patients dosed at the 10 mg QID dose of oral insulin was statistically significantly higher than in the 36 placebo patients (p=0.0368).

i read this wrong
icon url

urche

10/30/06 2:27 PM

#36546 RE: drbio45 #36513

EMIS

Multiple posters here today have been eulogizing EMIS following the results released today. I still don't know what to make of the results because the PR is just SO BAD that it makes a shameful mockery of the company. The market seems to agree, punishing the stock by 25% currently.

On the surface, the results seem promising, in that the high dose insulin group did show a signif decrease in HbA1c. That implies insulin was absorbed in meaningful amounts when 40 mg daily was administered. I'm willing to acknowledge that a phase II study doesn't need to necessarily show statistical significance. But, how are we to judge if they don't even release the raw numbers or P-values for the other groups?

Furthermore, the claim that oral insulin does not cause hypoglycemia is preposterous based on the limited info presented.

The study also demonstrated it was possible to deliver a fixed dose of insulin without the need to frequently monitor the patients' blood glucose or to titrate the dose for any reason. Blood glucose was measured on a monthly basis. ......There were no hypoglycemic events in any of the high responder patients as measured by HbA1c.

If the study really only measured blood glucose monthly, then of course the chance of finding hypoglycemia would be laughably low. Also, such a study design would miss a cheap and valid way to measure efficacy. Finally, HbA1c levels have almost nothing to do with hypoglycemia.

Ironically, there may be some favorable data and a drug that works here. And thanks to this poorly written PR, anyone bold enough to believe it can buy EMIS for a bargain today.

Urche