You may remember a two-part article I wrote last January on the very complex issue of whether or not a particular H1N1 vaccine might have caused new cases of narcolepsy.

Toward the end of the second part, I wrote the following two paragraphs:

“The group which recently reported on this hypocretin-targeting white blood cell tested the H1N1 virus, to see if the same blood cell that recognizes hypocretin as ‘bad’ would act similarly for parts of the H1N1 influenza virus itself. Lo and behold, they did find that part of the virus triggered a similar reaction in this highly specialized cell. This is one more piece of evidence pointing in the direction that infectious agents may trigger the onset of narcolepsy in genetically-predisposed individuals.

The Nature News piece that summarized these recent findings had a great quote from Gert Lammers, a neurologist working at Leiden University Medical Center in the Netherlands: ‘The results are very important, but they need to do a replication study in a large group of patients and controls.’⁷ Indeed, while this is all very exciting, replication in science is key. When one paper demonstrates that A leads to B, the public should say, ‘Hmm, interesting.’ When four independent papers demonstrate the same thing, it’s time to pop the cork on the champagne bottle.”

Wouldn’t you know?

The paper whose results I was quoting above has just been retracted by the journal which published it, Science Translational Medicine.

From a Nature blog on the retraction:

“But on 31 July, the authors announced that they have been unable to repeat a key finding: that immune cells from people with narcolepsy respond to hypocretin more so than immune cells from people who do not have narcolepsy. ‘Because the validity of the conclusions reported in the study cannot be confirmed, we are retracting the article,’ the team wrote.”

You can read the full post here.

This should serve as another reminder that one paper showing a correlation is interesting. Four independent papers demonstrating the same correlation, that’s something.

Replication is key.