doi:10.1038/nm0403-376b
April 2003 Volume 9 Number 4 p 376
Skeptical scientists skewer VaxGen statistics
Myrna E.
Watanabe
New York
When California-based VaxGen announced in February that its HIV vaccine
was ineffective in a majority of trial participants, few researchers in the
field were surprised. "All the animal studies pointed that way and even the
phase 2 trials...suggested that it wouldn't work outand it didn't," says
Dennis Burton (see
page 380), professor of immunology at Scripps Research Institute.
The vaccine, which was directed against the gp120 envelope protein,
elicited neither neutralizing antibodies (see
page 380) nor a cellular immune response, Burton notes. "The science
community is pretty angry by now because I think it was a pretty clear
failure," he says.
But with those results, VaxGen also revealed the tantalizing possibility
that the vaccine conferred 78% protection in African-Americans and 67%
protection in a group composed of African-Americans, Asians and other
minorities. Skeptical scientists were ready with sharp pencils.
Bette Korber of Los Alamos National Laboratory questions how VaxGen
derived the P (significance) value for African-Americans and the
grouped minorities. "If you move the African [Americans] out, you're left
with Asians and 'others' and [the results are] not significant," she says.
In response to press reports questioning its statistical analysis, the
company issued a statement saying, "the results...remain accurate as stated,
and the analysis continues." But Jim Key, VaxGen's director of
communications, now says the data have not been adjusted for multi-group
analysis.
Key says the next step is to find a biological explanation for the mixed
results. Asked why the company went public with incomplete data analysis,
Key says that VaxGen, a publicly traded company, was in a difficult
position. Once the data were initially unblinded, he says, the company had
to protect the information from leaking out for fear it would influence
stock trading. The initial announcement of trial results was followed by an
immediate drop in VaxGen's share price from about $10 to around $4.
Is it possible that a vaccine could selectively protect
African-Americans? Human leukocyte antigen (HLA) genotypes could certainly
affect both the transmission and cure of certain diseases, says Keith
Crawford, director of clinical research at Howard University's School of
Pharmacy. But Crawford and others are reserving further judgment until they
see a detailed analysis. "[VaxGen] should tell people what they've done,"
says Korber. "They should explain it to us."