In an astonishing move, AFFiRiS issued a press release last week touting AD04 as "the first drug ever to demonstrate clinical and biomarker effects consistent with disease modification in Alzheimer patients."

What is AD04? It's the placebo used in a phase 2 clinical trial AFFiRiS ran to test its actual drug AD02. At different dosages and formulations, 24% to 31% of the patients taking AD02 showed cognitive/functional stabilization or improvement, compared to 47% of patients in one of the formulations used in the group that got placebo, now called AD04.

Admittedly, AD04 isn't a typical sugar pill placebo, so the observation is somewhat believable. AD02 is a peptide-based vaccine that contains an adjuvant to stimulate the immune system to make antibodies to the peptide. AFFiRiS used the adjuvant, which is now called AD04, but not the peptide as the placebo.

Bristol-Myers Squibb (NYSE:BMY) ran into an equally confounding issue when it was testing its melanoma treatment Yervoy. The company tested Yervoy in combination with a peptide called gp100 and an incomplete Freund's adjuvant, but it turned out that Yervoy alone produced just as good of a result as the combination.

AFFiRiS didn't disclose what the adjuvant was, but it should be noted that GlaxoSmithKline (NYSE:GSK) licensed the AD02, but eventually returned the rights to the drug to AFFiRiS. GlaxoSmithKline has developed a variety of adjuvant systems. AS03 is used in the company's pandemic flu vaccines, AS04 is used in GlaxoSmithKline's cervical cancer vaccine Cervarix, and AS15 is used in its cancer therapeutic vaccines.

And Glaxo, with researchers in Université Laval, Centre Hospitalier Universitaire de Québec, have shown that an adjuvant called monophosphoryl lipid A can stimulate the immune system to decrease the plaques in brains of mice .

So AD04 is a wonder drug?
Not so fast.

I'll be the first to admit that sometimes drugs are discovered by accident. Merck's (NYSE:MRK) Propecia, for instance, was first tested as a treatment for benign prostatic hyperplasia before researchers noticed that bald patients were regrowing hair. It's certainly possible that the adjuvant helps Alzheimer's disease patients.

The problem here is that we don't have anything to compare the seemingly good results to. It doesn't make sense to call AD02 a placebo now. At the very best, this is a single-arm trial on patients with early Alzheimer's disease, which no one in their right mind would set out to run. Alzheimer's disease endpoints are too subjective and the decline in patient cognition and function is too variable to be compared to historical averages.

And, you'll recall from your sixth grade science fair project, science is hypothesis driven. You come up with a hypothesis, you test it, and see if your hypothesis was correct.

Unfortunately this is nothing more than an observation, albeit a highly interesting one.

Just like Merck had to do with Propecia, AFFiRiS will have to go back and run a prospective study with the hypothesis that AD04 works better than a placebo.

Let's just hope the company picks an inactive placebo this time.