Meningococcal Trial: some reason for hope
With a biased view as an investor - albeit with only a small investment at stake - I do care about the potential of XOMA's BPI Platform in infectious diseases. As I already pointed out, I will take any results from a severe meningococcal sepsis study only as a try to prove a concept.
The accompanying editorial in the Lancet (Marcel van Deuren, Petter Brandtzaeg) gives some background information and additional analysis on the study. If I look at the available data more closely - and in a way it is only allowed to try to get some more insight on a possible benefit that was missed with the study design - I can find some encouraging signs:
1.First it seems to me some reiteration is needed on the problem of the death rate in the placebo group: of 203 patients enrolled into the study with placebo only 20 (9·9%) patient died. These are about 40 patient less than what would be expected of the 30% mortality in the group with fulminant meningococcal septicaemia.
The authors state, that The trial was therefore underpowered to detect significant differences in mortality. Was it - as the authors of the study suggest - because most deaths occurred in the interval between identification of patients and study drug administration? I doubt it. So, what went wrong?
2. Some albeit hypothetical reasoning adding to my point:
A. Of 1287 patients (no mentioning that all of them had invasive meningococcal disease) screened, 57 patients were excluded (because they died or who met criteria for imminent death before receiving the study drug), another 34 of the patients enrolled died. Overall this gives a total of 91 deaths = 7.53%. Expected death rate with overall invasive meningococcal disease is 10%. Based on the expected death rate of 10% the patients screened with invasive meningococcal disease might have been 1208 only.
I conclude from my hypothetical calculation: A minimum of 1208 patient were screened with invasive meningococcal disease. From this number 1/3 would be expected to have fulminant invasive meningococcal disease = 402 patients.
The overall screened patient population however counts a total of 450 patients with fulminant invasive meningococcal disease: 190 in the rBPI21 group, 203 in the placebo group, 57 patients that were excluded. Actually this is more than expected statistically. The disparity might be due to wrong statistics I use, or a bias in the referring procedure leading to a more severely ill patient population that was screened overall. Detailed analysis of the study might give more insight.
If we look at it from the other way: 1287 patient were screened and 892 patients excluded gives 395 patients (=the effective number of patients enrolled into the study with fulminant invasive meningococcal disease). Note: this is close to the 402 patients expected statistically.
The point in mind: the 395 patient studied are close to the expected number of patients with fulminant meningococcal disease and should have a death rate of closer to 30% than to 10%. But they do not if we look at the placebe group with a death rate of 9.9%. The conclusion that the trial (was) underpowered to detect differences in mortality does imply that the low death rate in the placebo group is due to the high death rate prior to beginn of study drug administration.
For further speculation therefore I assume – to be on the safe side with my calculations – that of the total of the screened patients only 1208 patients had invasive meningococcal diseases. (More proper results might be given in the result section of the Lancet. If not: the question arises whether something went wrong with the selection of the patients?)
B.Based on the 1208 patients, statistically, a overall mortality of 120 patients (10%) would have been expected (or even 128 death if we calculated from 1287 patients). From all 1287 patients screened only 91 patients died. Statistically, I would have expected at least 29 more patients to die. This disparity in statistically expected deaths versus reported deaths in the overall screened patient population indicates that not the underpowered trial patients are the problem but the recruitment of the patients.
Or, could it have been that all 29 more patients expected to die (treated in the rBPI21 group only of course) did benefit from the treatment? If from the 190 patients in the rBPI21 treated group not 14 but 42 patients had died untreated ... Now, the low death rate of 9.9% in the placebo group reject such a simple assumption
C.Just to go on with some speculations: Since (according to the editorial in the Lancet) the expected 30% mortality in the group with fulminant meningococcal septicaemia makes up the bulk of the 10% mortality it could be calculated that the 91 patient that died were part of a group of 273 patients with fulminant invasive meningococcal disease.
Personally I would have preferred if this number would come closer to the 395 patients as reported in the study (or the 450 patients as calculated above).
I think the question whether something went wrong and in case which part of the study was messed up merits some further investigation.
The bad news still is valid: that a beneficial effect has not yet been proven with rBPI21. I am however not so pessimistic about the potential of the drug. I take from the available summary that at least the full report has to be read for more insight.
Regards,
Manfred |
