Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
ICPT
AUPH
WRT Repatha trials:
If the trial is huge for efficacy reasons, it is likely that the Number Needed to Treat (NNT), even for Relative Risk Reduction (RRR), is poor. And for Absolute Risk Reduction (ARR) it is even lower. This is the general flavor of the post you linked to - and it isn't wrong, but I would suggest it isn't exactly right either.
The repatha trials are just a continuation of Big Pharma's attempt to get the maximum patient population - i.e. the one that just finished was a fairly indiscriminate, almost all comers, trial (I doubt there are too many secondary patients that didn't meet the enrollment criteria under existing SOC).
But the risk is not flat across the population - there are at least 3 or 4 biomarkers that probably indicate much larger risk. E.g. High Apo(a), ... .
So there is still substantial possibility to get much better NNTs in cardiology - we just need to have very clear ideas of risk biomarkers, and targets other than LDL. The problem is that there needs to be an incentive to target them - but in a universe where there are no limits to $ spent to treat one person in a population, it makes sense to target the biggest possible population by including as many nonsensical patients as possible.
IONS
AUPH
What the heck!?! $AUPH pic.twitter.com/duZtu8mTar
— Bio Stocks™ (@BioStocks) January 28, 2017
AUPH
AUPH
AUPH
AUPH
$IONS
Scary misconceptions of potential FDA Commissioners – Tyranny of Statistics:
Recently there have surfaced multiple candidates for head of the FDA that very obviously did not understand even the most rudimentary facts about drug trials. Some of these misunderstandings are, perhaps, understandable, because they involve complex and detailed knowledge about diseases and drug history. E.g. compared to engineering design, the failure rate for drugs is much higher between conception and final approved product. But some misunderstandings are egregious because they show ignorance of fundamental laws of the universe. And perhaps the most egregious of these is what I refer to as the Tyranny of Statistics.
More completely, some of the various misconceptions that have shown up over the last few months have been:
a) The belief that observational data will suffice to prove efficacy. Perhaps it will if the efficacy is large enough. But on this board we should all know that for more typical efficacies (e.g. HR=0.6) even case controlled studies used for phase 2, only infrequently pan out in phase 3. (Case controlled studies are perhaps the most powerful version of an observational study. Yet for small efficacy they are wrong more often than right.)
b) The belief that having safety trials only will somehow save money and time – when, in fact, for ‘chronic’ diseases it is normally the safety that drives the trial sizing and, thus, the cost. In other words, this problem becomes a #c problem (see next item).
c) But the most egregious piece of ignorance on display was quoting the time it took to get insulin from invention to patients (2 years was the number given) - as if that is comparable to the time it takes to get most modern drugs through the process. This, very literally, is as bad as the infamous NY Times editorial from ~100 years ago that said rockets would never work in space because there is nothing to push against. This comparison to insulin is an egregious and fundamental misunderstanding of fundamental laws of math - what I variously refer to as either the “Tyranny of Statistics” or the “Biotech Version of Shannon’s Law” (Shannon’s Law, for non-engineers, is a fundamental law about how much information you can get within a noisy system.).
Below is a graph of that illustrates the Tyranny of Statistics. It shows how trial cost (I use man-years on trial as a proxy) changes vs the HR for which you need Statistical Significance. There are various curves shown, each of which is for a different background rate of events in the population – because, obviously, the rarer the event in the untreated patients, the more patients you need to enroll to get the required number of events (Note that I represent the frequency of events in untreated patients by the half-life in a placebo population – i.e. how many years until ½ of the placebo group has had the event.).
Finally, the assumptions … There are many, and they shift the curves around a bit - but none affect the overall concept of the Tyranny of Statistics. The key assumption I used to generate these curves is that, from a cost perspective, I assumed that enrollment is the equivalent in cost to 1 year of patient care on the trial. Obviously this can be very different for different diseases and protocols. Other, lesser assumptions, include enrollment ratio (1:1), alpha=0.05, exponential event curves, no trial powering at all, … .
And, now – to actually read the curve. Insulin vs PCSK9 Outcomes:
1) Insulin – HR is undoubtedly at 0.1 or better even in the very early (1930’s) patients. And the rate at which new T1D patients die probably has a half-life near 0.25 years. So, reading from the curves, the total number of Man-Years needed in a trial would be around 10 Man-Years.
2) PCSK9 Outcomes – The trials are designed to be stat sig even with HR only 0.9. And even assuming that the background rate of events in the untreated group is 20% per year that gives a half-life of about 3 years. So the total number of Man-Years needed in a trial would be around 11,500 Man-Years.
PCSK9 trials will cost more than 1,000x as much and undoubtedly take much much longer – perhaps 5 to 10 times as long to accrue and wait for events. This is an immutable statistical fact – and, literally, the only ways to adjust it is to either use a surrogate (with all the risk that it doesn’t end up predicting actual benefit), or enriching the populations to increase the event rate.
Overview conclusions and thoughts:
A) As more of our drugs become ones to treat chronic diseases (i.e. a low rate of events) and have only incremental efficacy (e.g. HR=0.8) the more expensive the trials will be. By a LOT(!).
B) The problem becomes even more obvious if you compare two separate paths to, say, an 80% reduction in events. Case 1: One miracle drug has HR=0.2 and the background rate of events has half-life of 6 months (24 Man-Years needed). Case 2: 6 drugs, each with incremental efficacy of HR~0.8, will get to about HR=0.2 when combined, but the cost for the first is 870 MY, the cost of the second is closer to 1000 MY, and the last is close to 1900 MY – for a total of around about 7000 MY. A total of about 300x more expensive for the same total clinical benefit.
IONS
IONS
IONS
IONS -
IONS - Notes from Ionis 1/5/2017 Pipeline Update:
Before starting – note that with Ionis you have to listen to the conference calls. Stan will clarify things that the company never puts in writing. Also note that, where possible, I have broken the data below up by drug (as opposed to whether it was in the presentation or in the Q&A). Finally note that I am assuming that anyone reading these notes knows the overview of the pipeline so I am not recapping this.
Volanesorsen data:
The big surprise is that they showed a stat sig benefit in pancreatitis. (Comment: undoubtedly pancreatitis is the #1 morbidity in hypertrigiceridemia. However it was a surprise that it shows up as stat sig in a trial of only about 55 man-years and 2:1 enrollment to boot. But this then begs a question – what definition of pancreatitis did they use?). It showed up in both the “non-FCS” patients (n=106) and in the FCS patients (n=7).
Random note: they clarified that some of the “non-FCS” patients could actually be undiagnosed FCS.
In answer to a question they said they will submit the data from this generic high trig trial to FDA (of course), but will not be pushing for this population to be on the label because it would require a much larger safety database. (personal prediction – regardless of whether it is on the label, I predict there will be off label use, maybe even as many as FCS patients because there are many more high trig patients than FCS patients. And they have similar pancreatitis issues.)
LDL – in answer to a question they said that the treated population did have an increase in LDL, but that high trig patients have very low baseline LDL (average was given as around 60 in this trial – and about 20 in FCS patients generally). And if the rise is enough the patients can go on a statin.
(Comment: I’d hoped to also see some data on diabetes since high trig patients generally have glucose metabolism issues. But I’d expect in the published paper (of course).)
Base Trig levels were not given in the charts – but it is possible to reverse engineer them to look for imbalances. (treated was about 1220, placebo was about 1360. And note that once the FCS patients are removed the remaining 106 patients have about a 1200 – still very very high.)
FCS natural history data – they made reference to publishing a paper on this topic soon. (They have publicized such data before in poster form – although I can’t find it in my files)
Platelet data:
In answer to a question on how Ionis uses the terms “significant” declines and “serious” declines they said they use the standard definitions (although they made reference to their being more than one definition). This question was clearly asked in reference to, for instance, AF’s tweet that IR told him they were using a different definition (i.e. serious = Grade >=3 AND bleeding).
As they do every cc, they clarified that they have seen no serious platelet declines since the ones that triggered the cc in mid-2015.
They will be publishing a paper on all their completed trials and platelets – the paper is due to come out very soon.
DMPK
Drug was not potent enough (reminder – RNA drugs, as a class, have a very difficult time getting into muscle) so they will be halting this drug, and waiting for a much better pre-clinical asset to be available. The pre-clinical asset is both a more potent sequence (said about 3x more potent in its ability to interfere with protein creation) and using muscle-LICA (about 5x more potent). In combo not quite the benefit see for liver-LICA, but still not bad.
FXI-RX
Brief clinical summary: clotting in the dialysis equipment was 40 to 60 percent less than placebo (both 200 and 300 mg doses did about the same). Bleeding for 200 mg dose was about the same as placebo, but 300 mg dose has 3x the bleeding (albeit mostly minor). (Comment: given the small size of the trial that just finished and the fact that the drug has site specific effects, I suspect that they will have to carry forward both the 200 and 300 mg doses at least into the next, larger, trials in order to understand the trade of bleeds vs clotting in whatever indications they pursue.)
Bayer decision is due in Q1 but hasn’t been made yet. (someone on twitter said Bayer had returned the drug – but that was not what was said on the cc)
GCGR-Rx
They are clearly pushing this - Stan chimed in at the end of the Q&A session, when there were no questions about the drug, to re-iterate his belief that there is a place for it. (Comment - I can see, best case, a possibility of a small niche somewhere, but don't think it is likely to be large. Too weak a drug, the 3rd line is too fragmented. and there is still some liver risk that only shows up infrequently (i.e. large trials))
Stan noted that high doses of metformin would blunt the efficacy of the GCGR drug. (Comment - ok, but given that metformin as first line is about the only fully standardized thing in T2D, that doesn't seem imply that in later trials it will be more useful. Especially since they didn't run the trial in metformin intollerant patients.)
BTW – Random note: I’d be very surprised if the company isn’t tracking social media comments (including this site) since many of their comments appear to be in direct response to specific critiques.
GNCA
IONS - Had time to vet my numbers (e.g. did a simple exponential integrator to validate my AUC calcs). All correct except the simplest number (but that number doesn't effect the other numbers).
The correction: the nusinersen Cmax in the Cerebrospinal Fluid (CSF) is only 40x the serum Cmax for Drisapersen or Volanesorsen. I.e. As I noted before they have really ramped up the CSF concentration to get oligo uptake in the neurons - but *only* 40x, not 40,000x (I obviously slipped a mil, micro or nano somewhere).
So - some potential explanations are:
a) There is no reasonable serum exposure level for RNA delivery technology.
b) When designing nusinersen IONS ignored normal rules about designing/testing 2MOE gapmers because they knew serum exposure would be much lower.
c) SMA does, in fact, induce thrombo and renal as the kids survive longer.
I tend not to believe #a for common sense reasons combined with the fact that thrombo is a well known issue with the technology, and there is reference to side effects being at least somewhat linear with dose. But both #b and #c seem plausible to me. Certainly IONS is generally cavalier about pushing safety - it is the inevitable consequence of their discounting of safety concerns. But #c is also plausible since it is well known that SMA kids will have chronic infections (of the lung) and have systemic sequelae (even though they are normally of *much* less concern than central weakness).
How to differentiate:
a) timing data - e.g. when does thrombo data show up in treated kids vs untreated. Some of this may show up in the FDA Review documents (which aren't available yet).
b) animal tox data - but I don't know that we'll ever see this at the necessary level of detail.
Comments on the BIIB/IONS approval:
a) thrombo showed up on the label - 11% Level 1 or 2, 0% in placebo arm. Note that this is a much lower dose (12 mg) than in ApoC3-Rx or TTR-Rx and is intrathecal.
b) This is classic Stan Crooke - they were silent on this prior to the label release. Note a correction to my twitter timeline response (I deleted it, but nonetheless) - IONS wouldn't have had the unblinded data for this SMA data at the time of the FCS/TTR thrombo announcement since that announcement was in late May, and the SMA trial wasn't announced until August 1. Nonetheless they knew, unambiguously, that Thrombo is clearly a concern and were silent about what they saw in the SMA trials.
c) It was approved for all ages, including adults, with no apparent restriction on SMA Type (1, 2 or 3).
d) IONS gets another $60M milestone (I presume in this year) thus making me incorrect in my prediction that this would be the first year in 4 or 5 with a significant reduction in revenue (almost $300M last year).
e) proteinurea also shows up on the label. This is a well known side effect of this class of drug - and it shows up as a trend in the data quoted in the RCT label data. There is also mention of older patients seeing it - but it isn't clearly described in the label so I'll wait until I can read the statistical and medical reviews.
CLLS BLCM
I am a little worried that I sound like a broken record, but I would suggest that, broadly, the risk to CLLS strategy is that their response rate and/or mortality rate will get substantively worse when they roll it out more broadly (i.e. to more trial centers, and even more patients within the same centers - so less time/care with each patient). I don't track CLLS routinely, but for example @BiotechBbqBeats notes after the first 2 pts, they have only 2 of 5 patients getting to MRD negative. And another way to think about this is to compare these early results to the first data from auto CAR-T.
More narrowly, again as I've noted before, the problem that CLLS hates to acknowledge is that they need to lympho-deplete MUCH more heavily than auto CAR-T. That means much more expense and likely higher mortality in order to get the same response. But I wouldn't expect this to really show up until they start rolling out to a lot of trial centers.
Finally, FWIW, I think they are really more of a competitor to BLCM than JUNO/KITE because, like BLCM, they require the lympho-depletion.
AKBA and GSK HIF drugs
IONS ALNY
IONS
IONS – The bear case.
This will be the first of a long running series of bear case posts on a variety of different stocks. As a general rule I will only be writing up those companies that, if the bear case is addressed, have something left. I.e. I will explicitly exclude companies like PPHM or NWBO. Note that at the bottom of each bear case I will also write up a simple bull case as context – but the emphasis is on the bear. My goal is to make the bull vs bear debate more substantive.
I picked IONS primarily because, although there are many IONS bears, the bears’ case is almost reflexively focused on side effects. I’d find it much more useful if their bear arguments were less reflexive. As with most of biotech there is plenty to be bearish about without falling back on reflexivity.
1) Valuation - $5B is a lot for a company with $300M revenues last year (and probably about $200M this year)
a. Most of the drugs in their pipeline are partnered (royalties at about 15%) and most are orphans – i.e. smaller patient populations. Even if, optimistically, 30% are eventually approved it won’t equate to more than $1.5B/Yr and that is 7 or 8 years from now (with a heavy risk discount back to current stock price). (The bull case is that the pipeline is 10’s of drugs – and they have a history of getting large milestone payments)
b. The wholly owned drugs are much riskier than the partnered drugs – e.g. volanesorsen (for high trig) is likely to follow the same trajectory as other lipid modifiers (i.e. low sales $). (The bull case here is that FCS is not the same as generic high trig)
2) Drug specific TTR-Rx (partner GSK):
a. Watch issue: Once both IONS and ALNY TTR polyneuropathy results come out the IONS results are meaningfully worse from an efficacy standpoint
b. IONS thrombo data in TTR Poly is bad enough that it more than offsets the much more cumbersome and toxic ALNY infusion system.
3) Drug specific Volanesorsen:
a. It is only a lipid modifier – and sales of lipid modifiers are anemic without any clinical efficacy of benefit to the patient.
b. The thrombo data is much worse than has been presented by the company (see below for details of what the company has said and not said – but the odd thing about the current bear position is that much of it is in actual contradiction to what the company has explicitly said, and that seems odd).
4) Drug specific FXI-Rx (partner Bayer – with royalties greater than typical for IONS):
a. No specific results released after the dialysis trial and real release (first real release expected early Jan cc) will show no benefit of bleeding vs coag. Or worse, show adverse events (e.g. even a small move of platelet count could be bad given the indication being pursued).
5) Akcea drugs (lipid and metabolic targets via liver) are all largely pipe dreams:
a. NASH etc are, at best, poorly understood and with poor surrogates so the chances of success are small.
b. The side effects of the HbA1c drugs are likely to take a LOT of money to resolve – and several of their drugs in this area actually had big enough issues that they were visible even in very small trials (e.g. liver enzyme issues).
6) Thrombo :
a. Thrombo is an on-going risk in TTR and FCS and even though all the data to-date says it only occurs early on (within first 4 months of dosing) it may re-appear much later (in Drisapersen, a related technology, the thrombo was more severe, but only showed up after several years).
b. In the pursuit of non-liver indications (e.g. DMPK) they up the dose above 300 mg and risk much more extensive thrombo that dampens all the rest of their pipeline.
Brief bull case (largely for context):
a) FCS may not be ‘just another lipid levels’ disease. The literature on FCS is sparse, but there are hints that, for instance, even for a given trig level the FCS patient has a much greater chance of pancreatitis (chronic or acute). BTW – in my previous write up of IONS (almost 1 year ago) I speculated that the FCS primary endpoint was likely to have success, but less likely than the SMA trial. I now change that to the chance of success at the primary endpoint is just as high as the chance of success in SMA was 1 year ago. The basis of my change is a re-evaluation of the ph2 results (2 of 3 FCS patients went well under 500 – and that is very uncommon in FCS), plus the FCS subgroup in the just announced high trig ph3 RCT.
b) TTR-Rx results may show very substantive benefit in the cardiomyopathy subgroup – and if so it is likely to also have benefit in wt TTR cardiomyopathy (a huge market – although one that will require longer trials and rollout of new tech and SOC to differentiate TTR cardio from other amyloidosis cardio)
c) DPMK results (planned to be discussed in the Jan cc) show clinical efficacy even in the short and small trial.
d) FXI-Rx results in dialysis reprise the earlier ph2 results of large clot reduction with only mild bleed increase. Complete clarity by IONS on different types of bleeds and clots and severity will be required here – so this is definitely a long-shot given IONS history about being fuzzy when it comes to safety.
Thrombo on the record - Finally, a brief summary of what IONS has *explicitly* said about thrombo in their TTR and Volanesorsen trials:
a) All such thrombo events have been transient. And all occurred on or before around 4 months of dosing. And subsequently, including as recently as last week, they have said no such thrombo events have occurred since the June disclosure (i.e. after that 4 month mark).
b) The thrombo is only occurring in diseases which already have platelet anomalies – e.g. they explicitly note that there have been no thrombo events in volan non-FCS patients despite many more such patients.
c) None of the thrombo events have resulted in significant bleeds. And all the thrombo events have been transient.
d) Note that I do not believe they have been explicit about the exact characteristics of the thrombo – exactly how severe nor how long nor exactly how many patients (other than to say ‘less than a handful’ across TTR and FCS)