A Few Cathartic Thoughts On the Disconnect: PR vs. Analysts
Was easy for the analysts to not overreact after talking with CMO/CEO because most had nothing in there for oncology and now they can cut a lot of RD and SGA and see revenue growth drop to the bottom line. Those would be value enhancing for them. They got more meat around the decision from mgmt and have their same view -it’s a risky pipeline asset they’ve ignored, so a change in P1 doesn’t change their views much. Some of us and some of the sellers had way more and see the path as either too far off or non-existent now. Both are now meeting at the same place despite starting out with divergent views
Mgmt will have to provide investors more when they discuss the new protocol. They should use that opportunity to discuss what the wider cancer team saw and why they did this and what supports their thinking it can increase efficacy so that it outweighs or balances the SAEs seen. A KOL talk reboot. Maybe it helps reset the plate for some new risk-takers or cancer specialists to give it a look.
Future Enrollment Issues
All those oncologists are comfortable with trial dosage changes early on given so much of all this is really guesswork anyway. I don’t think that will turn them off to the trial in a way to make it much harder. From my clinical site research a month ago, there’s countervailing factors that probably won’t change the ability to get the new patients enrolled, albeit slowly again. There’s a lot of 1L-3L trials but very few trials I saw specifying highly resistant, late stage tumor survivors so that helps them a bit. The patient exclusions or tumors they may choose to focus on will likely shrink it some for certain cancers, like melanoma, I assume. But that’s good too for increasing probability of success.
Our collective Whiplash from the new analysts to new revisions. A clue?
I have to say that I’ve never suffered such a crash of optimism from my giddiness of finally seeing an analyst value Sort1+ to the need to change the protocol. I think that’s a clue to answer a few of these questions around how/why at this time given those analysts just published.
My interpretation is that discussing a change probably came from the outside investigators (or PPD/Rothenberg) who think there's a better way to boost efficacy while keeping SAEs tolerable. After all, those are the only people in this equation who have the large experiential sample to know how these things happen and how you might want to change it. Marsolais does not have that large analog sample toolkit like the investigators all do so he wouldn’t have thought to do that. Is it a surprise that they immediately talked about the Cybrexa approach, where it’s a PDC and at MD Anderson? I don’t think so. THTX probably would have done as Qwerty thought, just keep cranking away and you’ll get 2 out of 10 somewhere and move on. Investigators probably saw it would be beneficial to try to boost efficacy now if you want to get to P2 on 40-70 patients. That’s why we never heard a thing from CEO/CMO except the superlatives based on pre-clinical and 18 patients in 1a, and a handful who had some efficacy, just not great efficacy. They need great. The strong background science, which investigators also all bought in to, and the less potent data was enough to urge THTX to change the approach to try to optimize for more of those Prostate guys and less of the “meh”.
How I’m Thinking About Some of this Now, for what it’s worth.
First off, it really shouldn't be a surprise there could be changes to any drug trial very early on. Clearly underestimated by a lot of us despite seeing some signs. Trodelvy had significant safety related changes in P2. How often do changes result in positive results coming forth? I don't know, but I do know we have to up the risk factor a chunk and that lowers value. The trial is now re-risked, not de-risked. That's an objective interpretation and I think they and the analysts would all agree with that. It's just those analysts always had it risked higher so they aren’t moving much.
I know you think I'm whistling past the graveyard here, but I'm trying to be constructive given there is some valid science behind it, but it's worked in a spotty way to thwart any clear POC. I think if there were dreadful results of literally not seeing any kind of “activity signs” and just SAEs, they and the investigators would stop the trial at this point. I guess they’re telling the analysts and us it’s not there, it’s early still, and there’s reason to believe we can still try to get to that low hurdle on POC -2 out of 10 responses for a cancer type in a new regime. No doubt the PR and perhaps reality doesn’t look as promising as pre-clinical once did. The corollary of the PR would be that everyone involved did see enough activity signals and ability to deal with SAEs to believe it is worth revisiting dosing early on instead of stopping the trial altogether.
Those Katana lab guys and Marsolais will be working overtime to seek any clues. It could take time. But they are the guys who know what they have, are a leader in understanding sortilin in the body and peptide design and how the PDC should work. The investigators also know a lot. Some science questions may never be answered, but they can probably answer some and tilt probabilities a tiny bit toward efficacy from all the new data or will at least try. Will that be enough to get 2 out of 10 and a POC? Guess will eventually find that out, but few will bet on that now. Paul is doing the smart thing as CEO to restructure the company to withstand the news—cut costs a lot and ramp revenues best you can.
Ended up not being a 1B
One way to objectively look at where the program stands this weekend is that it turned out it wasn't really a 1B basket trial. The 1A recommended dose and schedule, while informed by the initial 12 patients, was basically a statistically thin result from running 3 patients for 6 weeks twice at 300mg to see if they lived. It didn’t stand up as the best dose when put to the test. The Leeds analyst states it well:
“In many circumstances, dosing regimens for any Phase I-stage experimental cancer therapies is an educated guess based on preclinical studies, and pharmacokinetic data. Such studies can often fail to be perfectly superimposable onto a clinical background. This appears to be the case with TH1902.”
If you read the FDA Phase 1 stuff, it doesn’t rule out you won’t go back and revisit it after you learn some things. It’s part of the process they demand. The PR is saying a standard 300mg dose definitely wasn’t robustly effective after what we’ve seen in the trial so far. Because of fast-track, I’m sure the FDA was seeing all this and part of some of these debates. They may also be helpful through this to some extent or coud provide industry insight contacts.
Maybe it can be improved upon with the patient selection things they learned and in dosing methods that others have used to improve their results with PDCs so that it increases the chance some patients get to more powerful efficacy. There was some efficacy to TH1902 so some elements of the POC are working in some patients. How do you get more potent efficacy in more patients consistently? That’s CMOs job now. If they had the money, they probably could have run something larger in P1 and played around with things but they chose to look at more cancer types rather than to look at different dosing schedules. Cybrexa is doing both in their trial, as have others. Lutathera trialed two different doses I think.
It wouldn’t surprise me one bit if the 1b results looked a lot like 1a —tolerable SAEs but the efficacy issue didn’t really add much to the trial except questions. Recall that all of us even questioned why in 1A we didn’t hear more about the high sort1 cancers like TNBC, Ovarian, Endo, but heard about low ones like Prostate? It was odd. But it doesn’t negate that they have seen significant tumor regression and disease control in places, it just creates questions as to why there wasn’t a broader swath of patients seeing success like those few we know of. The MOA says it should do far better across the board. That’s the key question the investigators asked and what they need to analyze and get right to be commercial at this stage in the trial. We got excited about a few solid efficacy results. But it just wasn’t enough overall and didn’t see improvement as enrollment ramped.
It appears to me that you can somewhat skip over the robust efficacy question in 1A where you may just be happy to see any tumor activity, some durable responses and some overall survival benefits while keeping SAEs tolerable like what they did see. You hope to see really good disease control rates, but it won’t stop you taking that dose to trial it in the basket and to then focus more on disease control and hope it expands with the expanding enrollment. It’s all just Phase 1 anyway. The clearest message in the PR is through 1a and 1b, they got a mixed bag on disease control and responses such that in their view you better stop and reassess at this early stage if you want any chance to get the kind of efficacy numbers you need to graduate to Phase 2. It’s a program black mark for sure since there was not one element of pre-clinical that was short of stellar success, and their issue is whether they can overcome that through science or not.
Now they have 26 or so patients with that 1A dose to determine what’s the most optimal dose, timing, type of patient exclusions, etc… They’re back to 1a experimenting, it ended up not being the 1b basket. All they know now is that the max dose tolerable is still 300, but that dose has limited and mixed efficacy signals for reasons they can only guess at now. Secondly, to the extent the pesky SAEs are there, how can they be lowered so patients can stay on longer and maybe help boost efficacy? They also know that when they lowered from 420 to 300, they saw SAEs lower a lot. But I think the new objective is primarily how to tease out efficacy in a different way. They must see some data from pre-clinical and this trial to think 150 or 200mg every few days could help achieve those ends. The Cybrexa guys have all frequent, multi-day injections and not the one and done standard. I hope they talk and learn something from them or the investigators at MD Anderson on that trial (lead investigator is different).
So that’s where it is in my mind —they’re back at 1a with more data, more patients through, more thoughts about the drug in humans, all the "patient selection" stuff they learned. The big question is why not consistent, improving efficacy in at least 2-3 out of 10 patients. It sure seemed like a low bar, but that is not the correct way to think about it if you need to get solid data to the FDA to move on. You want to KNOW you can get there, not hope. They were definitely not seeing that and the efficacy question is the nut to crack and they need to help explain that better to us all and the FDA. Rewind the clock in essence, but now knowing robust efficacy is harder to get to than pre-clinical showed.
Like the science guys on this board said a long time ago, don’t forget that the tumor environment in humans is way different than in rats.
Some of the Questions I have
Investigators Know More?
It’s interesting in hindsight to think of what debates they may have had with investigators back in 1A that led to delays. I’m totally guessing but maybe part of that delay was investigators comparing TH1902 to other drugs and seeing efficacy robustness wouldn’t be enough to get you to Phase 2 and they were suggesting trying to do different dosing schedules like they probably see in other trials. But of course, that would have added costs and delays and THTX pushed back and was confident the robust pre-clinical would show through. Maybe investigators were more skeptical but saw enough disease control and response data to agree to move on for now. I’m sure there was some intense debate going on around that point, and now it’s come right back to it. In hindsight, you’d be right if you wanted to see more durable results before moving on and it should have been experimented with back then. They are there now anyway.
It’s sort of odd because before all this started they would talk about how the MOA could lend itself to a “new paradigm” in how to treat cancer. Then they’d talk about how this might allow doctors to use lower doses more frequently to keep tumor growth low, metastases low, and any side effects low. So they clearly had thought that might be one way to play around with it, but they didn’t have room to test that idea. Big mistake! They should have approached it that way.
Efficacy vs. smaller doses?
We’ve all asked the question how can you see more efficacy when you’re lowering the dose? They need to discuss this. The “normal” therapeutic dose is equivalent to 230mg of TH1902. So 100 would be 50% less, 150 would be 35% less and 200 about 12% less. Does this lose the advantage of a large dose? I’m sure they would say the internalization factor will still boost that a lot. It’s probably less in humans, but probably does have some concentrating factor of unknown level. They’d have to provide something to back that up to make it believable given where they are now. It may be that case, but pretty hard to tell. If it isn’t, then they’ll pretty quickly see that in a low dose/more frequent trial.
This kind of cellular level processes is way over my head and way too complex for me to understand much of it. There’s endless permutations both good and bad on how different dosing changes things. I’m sure there has to be some kind of analytical way to support how a more optimal dosing for a PDC or for docetaxal otherwise Cybrexa would have probably stuck with the standard one big dose every three weeks. They didn’t and there must be a reason. THTX needs to use whatever testing they did to figure it out too. They really need to use investigators, KOLs, the scientific big brains to think it all through. But there’s no reason to believe there might just be a “more optimal” way to do it since there’s industry precedent for what they’re embarking on now.
Lowering SAEs, what are they caused by?
On SAEs, they telegraphed in 1A that SAEs do lower with lower doses. But how do SAEs change with frequency? How much of the SAEs is due to immediate dose and how much to accumulated dosing? This is also where they may have some decent data from the 30 or so cases with data (my guess). The Leeds guy had questions around some of this and the sort1 targeting, some like JFMs prescreening.
“We were interested to see that one of the justifications for revisiting TH1902 dosing was that dose-limiting neurotoxicity and ocular toxicity were sustainably observed in many patients, and these are side effects intrinsic to microtubule-binding docetaxel as well. This suggests to us that TH1902’s pharmacokinetics may not be as specific for SORT1 receptor binding as originally assumed and this could be for several reasons that may require further foundational investigation. One possibility is that many of the patients enrolled in this initial 70-patient trial did not present with localized tumors that expressed sufficiently high SORT1 levels to support tumor-specific binding.”
“Neurotoxicity that is more commonly ascribed to docetaxel itself is surprising for TH1902, based on all preclinical data showing how SORT1-specific its pharmacology is. It would be somewhere between interesting and necessary to explore this possibility through immunohistochemical analysis of tumor biopsy samples in unresponsive patients or in patients exhibiting unusually intense neurological side effects. Another possibility of course is that SORT1 is more ubiquitously expressed in other non- diseased organs (and specifically in nerve tissue) and not solely in the tumor that TH1902 is targeting. This possibility is less amenable to analysis, unless TH1902 could be supplementally labeled to allow for non-invasive fluoroscopy or PET analysis to transpire. A third possibility is that the sortilin peptide to which docetaxel is bound in TH1902 is just not as specific for SORT1 as it needs to be for clinical utility. And a fourth possibility is that TH1902 is being cleaved into docetaxel and free peptide once administered intravenously in diseased subjects.”
From hearing Belivieu and Marsolais, it has sounded like they understand some of these issues and have tried whatever you can clinically to address it. Of course, humans introduce new complications and that’s what they’ve seen and need to try to best figure out. Some won’t ever be actually figured out at this stage or at all. Cells and tumors are still very mysterious. Guesswork is the best you can do and hope you tamper down the bad and amplify the good somehow. Just shows how complex all the moving parts are.
Big Brains Needed
Trying to go back and answer these question now isn’t easy and is a tough spot for Marsolais and Belivieu to be in. Maybe they have a lot of data that can really help and they have their Advisory Group and Investigator Group to provide input. I’m sure there’s plenty of people who can help them think it all through and they can only hope they get it right and meet that darn low hurdle - 2 of 10, then optimize more and get the 30% ORR in P2! Unfortunately, this setback has re-risked it and put in a harder hurdle for them to pass. It seems to work on some patients, so they need to figure out why or see if they can select enough of those types to create an effective therapy for that specific type.
Feel sorry for all of us and especially for these hard-to-treat patients who hopefully will benefit from something more “optimal”, but for now it’s back to the dosing blackboard. Quicker Paul can get to breakeven cash flow the better since modestly growing sales could be worth a multiple of 3-4-5x and he’d then have flexibility to raise money at those $300-500mkt numbers.
Hope it helps you think through some of the issues they and any investor face. I'm speculating on many things there but trying to keep it realistic.