This was written by an unknown person called Julian Bochardt in a epublication, which you can read if you like (CLICK).
It was interesting that Berger & Fox came to a similar conclusion and suggested that the risk of PML was nearer 1 in 44. Reassessing the risk of natalizumab-associated PML. Berger JR, Fox RJ. J Neurovirol. 2016;22:533-5.
Borchardt J, Berger JR.Re-evaluating the incidence of natalizumab-associated progressive multifocal leukoencephalopathy. Mult Scler Relat Disord. 2016 Jul;8:145-50. doi: 10.1016/j.msard.2016.03.005. Epub 2016.
METHODS: We analyzed post-marketing data about the incidence of PML on natalizumab, and quantified the risk by either applying the Kaplan-Meier estimator or, where this was not possible due to the unavailability of the respective raw data, using formulae yielding very similar figures.
RESULTS: In JCV-seropositive patients with prior immunosuppressant (IS) use, the incidence of PML during months 25-48 of natalizumab therapy is about 19.5 per thousand. Without prior IS use, the incidence during months 25-48 is approximately 7.4 per thousand, and during months 49-72, it is approximately 10.8 per thousand. If one additionally assumes that the JCV index is in the range 0.9-1.5, then the incidence during months 49-72 is around 6.2 per thousand in comparison to 17.0 per thousand when the JCV index exceeds 1.5.
CONCLUSIONS: Biogen’s statistics concerning the risk of PML on natalizumab, while in principle helpful, underestimate the true incidence systematically and significantly; realistic estimates of the long-term risk of PML are nearly double those previously published, with some patient groups carrying a risk that is almost nine times higher. Fortunately, a refined risk-stratification algorithm with the incorporation of such markers as L-selectin and CSF lipid-specific IgM bands has the potential to make natalizumab a considerably safer drug.
This eventually has prodded Biogen into a reaction (CLICK)
However, in the Words of Julian Brochardt “In a nutshell, they accepted my criticism as valid, but argued that, given the recent completion of four “large” clinical trials, estimates based on post-marketing data were de facto obsolete by now. I don’t quite agree with the latter”
Julian has a few comments on ProfG’s risk calculator too.