We have reached a repeat moment in the saga of everyone’s favorite reverse-merger pink sheet Coronacrapper, Cytodyn (CYDY). The time when, after a study failure, the criminal stock promoters, various shills, and the retail shareholder base of complete morons suddenly become “experts” in “teh science”, results be damned. [Catch up on Part 1 of But Teh Science Be So Gud, from August 2020, here.]
After three morning press releases and then last afternoon’s delusional CONference call rantings from The NaDDir*, and his Klown Krew, yesterday the company issued an 8k with some (and only some) actual data from the failed CD12 study of leronlimab (lovingly referred to as loserlimab at BuyersStrike! HQ). The filing is available here.
The document is almost more noteworthy for what it doesn’t include, rather than the various post-hoc data dredged subgroups that it does include. Nowhere does it actually answer the question, “Does loserlimab actually save lives versus placebo?” Not mixed with a variety of other drugs, not when mixed with a variety of other drugs and dexamethasone, but does it actually do anything head to head against placebo? In order to answer that question one of our overworked interns volunteered to back out the numbers from the scant data that was provided.
Table 1-1 provides some mortality data overall for the mITT population. “mITT” is jargon for “modified intent-to-treat” and typically mITT is frowned upon, especially when the data from the ITT (without modification) population is not given. Nevertheless, it is all we have from the company. From this we learn that there were 259 patients in the loserlimab arm and 125 in the placebo arm. There were 53 deaths in the loserlimab arm (20.5% mortality), and 27 in the placebo arm (21.6% mortality). Although not given by Cytodyn, the difference is nowhere near statistically significant, the p-value is a whopping 0.79.
Many of these patients were taking all sorts of other therapies, and The NaDDir* is especially proud of one particular post-hoc subgroup analysis, patients who were on other prior or concomitant treatments:
The given p-value is actually suspect as the company admits “Note: p-value is from logistic model adjusted for stratification factor and age based on non-missing observed data.” A more standard calculation of the p-value would yield p=.167, which is still crap, but at least not p=.79 crap. Behold the power of post-hoc data mining.
But this chart still doesn’t tell us anything about the effects of loserlimab alone versus placebo alone. How did patients who were only taking loserlimab do versus those only taking placebo? Does leronlimab really save lives? Cytodyn does not provide the answer. But we will.
In order to answer that question, one needs to back out the 309 subjects above, those who took other treatments, from the 384 total subjects in the trial. Then we need to back out the 58 deaths in that group from the 80 deaths overall. That gives us 75 subjects (21 placebo, 54 loserlimab) and 22 deaths (3 placebo and 19 loserlimab). Plug those numbers into Excel, and presto, the answer to the question, “Does loserlimab actually save any lives?” is a resounding “NO.”:
35% of subjects receiving loserlimab alone died compared to only 14% of controls. The p-value happens to be 0.075, results that, as The NaDDir* would say are “Very Exciting!” and “Approaching Statistical Significance!” So much for “saving teh livez” with “teh science that be so gud!”
* Spelled Thusly For A Double Dose of That Sweet Sweet Stock Pimping
THE CONTENT CONTAINED IN THIS BLOG REPRESENTS ONLY THE OPINIONS OF THE AUTHOR. THE AUTHOR MAY HOLD EITHER LONG OR SHORT POSITIONS IN SECURITIES OF VARIOUS COMPANIES DISCUSSED IN THE BLOG. THIS COMMENTARY IN NO WAY CONSTITUTES INVESTMENT ADVICE, AND SHOULD NEVER BE RELIED ON IN MAKING AN INVESTMENT DECISION, EVER. THIS BLOG IS NOT A SOLICITATION OF BUSINESS: ALL INQUIRIES WILL BE IGNORED. THE CONTENT HEREIN IS INTENDED SOLELY FOR THE ENTERTAINMENT OF THE READER, AND THE AUTHOR.