Just read the snarky tweets from Adam Feuerstein regarding HALO's presentation. Wow! The guy can be a real dick. I have generally enjoyed his articles and, even when disagreeing with him, have usually felt he was even-handed. He has to form an opinion or he won't be read, and with biotech, it is generally more profitable to be a nay-sayer. Good money is to be made at the expense of the overly optimistic. That said, it seems his tweets shoot from the hip and are of a lower quality than his articles. I hope he writes a piece on HALO and actually does a little research.
There doesn't seem to be much question PEG extends PFS in stage 4 pancan. The slide showing much better PFS event data in those with HA scores > 75 % is nice correlation. But that won't be nearly good enough if all PEG is doing is shrinking the HA coat and making tumor progression harder to spot. Survival MUST be prolonged and that is the key point driving all of the eye-rolling commentary from the likes of AF and Lowe; as if Torley didn't already know that.
AF and DL seem to imply that the combined Stage 1&2 OS KM curve is the most important slide and all the rest is mere distraction and an attempt at data mining. But data mining is when you do a post hoc analysis and select your end points after the data is in. Everyone (but them?), before the presentation, understood the Stage 1 data and the near impossibility the Stage 2 data could overcome the oddities found there. Really, only the Stage 2 data mattered for this presentation and HALO should be commended for laying it all out there, even if it confused those with only superficial knowledge of the company.
Unfortunately for HALO, the Stage 2 OS data has too low an N for the FDA to consider it sufficiently trending positive. And the fact the Stage 1 data can't really be discarded by them makes any dream about early "breakthrough" approval extremely unlikely. But that doesn't mean a good investor can't de-emphasize the Stage 1 data when handicapping the ph3 trial. Why? Once again:
- The AG control group for Stage 1 high HA showed a crazy high OS. After some large trials, it is generally accepted that the OS for Stage 4 pancan treated with standard AG chemo is under 9 months. Moreover, though less well established, it is generally accepted the OS is significantly less than that in those with high HA levels. So, how strange is it that the OS for AG in the Stage 1 high HA is 10.9 mos? This more weird than simple low N problems. In the Ph3 trial that established AG as one of two SOCs, (Von Hoff, NEJM 10/31/13) the AG cohort with N=431 had alive at months 30/33/36 and 39: 4/1/1/0 subjects. HALO in Stage 1, with N=21 had alive at months 30/33/36 and 39: 3/2/2/1 subjects. Their control group had a relative forty times the number of survivors at 33 and 36 months!
- The more realistic data of the Stage 2 cohort showing OS of 7.8 mos doesn't have a high enough N to overcome the Stage 1 weirdness. We'll need the higher numbers of a ph3. That said, all the numbers in Stage 2 lined up nicely with what you'd expect from the historical data.
- The OS and PFS of the Stage 1 high HA PAG cohort are almost certainly affected in a negative way by the TEEs, the minimum of three mos off PEG, and the 40% who did not resume PEG.
There will likely be a short attack on this stock in the next few weeks. AF and others will not know anything other than the face value top-line data (and will poo-poo anyone who looks deeper as delusional) and they can also be certain any Ph3 results are at least 12 mos away. But, for those investors, or companies, who can evaluate all of the information, this might present a buying opportunity.
