Last week, researchers published a preprint containing data about lenzilumab, a drug $HGEN is developing to prevent Covid patients from ending up on ventilators.

Preprint here: https://www.medrxiv.org/content/10.1101/2021.05.01.21256470v1.full.pdf

I’ve been getting questions and comments. So here are my thoughts.

1/10

Importantly, this paper includes positive data not in the company’s previous press release. The original analysis did not include 41 out of 520 patients who were randomized. An important question was how the analysis looked with those patients included.

It looks good!

2/10

Without including those 41 patients, those who received the drug had a 54% improvement in their odds of survival without needing a ventilator. With the full 520 patients included, there is a 90% improvement.

I’d view those results as essentially the same.

3/10

Still, if things swung the other way it would be something to worry about.

But the confidence intervals are very wide. For the 54% result, they range from 1.02 to 2.31; for 90%, they range from 1.02 to 3.52. P values of 0.041 and 0.043, respectively.

4/10

For reference, here’s my previous coverage:

https://www.statnews.com/2021/03/29/humanigen-says-it-drug-keeps-covid-patients-off-ventilators-but-are-missing-data-relevant/

https://www.statnews.com/2020/11/06/humanigen-touts-analysis-of-covid-drug-but-statisticians-say-disclosure-is-inappropriate/

5/10

With regard to the issues raised in those stories, of course this means that we have now seen both analyses and they agree. I still think the early disclosure of results was inappropriate, but that would only change FDA's decision if that disclosure impacted the data.

6/ 10

So does $HGEN get an EUA?

I don’t know. This is a positive study, but barely. A drug that keeps 6 out of every 100 patients treated off a ventilator would be valuable. (See table 3). But those confidence intervals are very wide.

7/10

It’s the kind of data set that could result in an EUA, but not if the FDA’s review reveals any other red flags.

So that’s still how I view these data. The lower bound of the confidence interval is 1.02, which would not be much of a benefit at all.

8/10

If I were an investor in $HGEN, or, for that matter, $RIGL, I’d worry a bit about $MRK experience with OncoImmune. $MRK bought OncoImmune because its drug helped hospitalized Covid patients in a randomized trial. The FDA asked for more data, and that stalled the project.

9/10

We don’t really know what the bar is for an EUA, and we know that it’s likely moving. That makes it tough to predict how the FDA will handle a result that just hit statistical significance.

In Wall-Street-speak, I would say I don’t have much conviction here.

10/10