Frequency Therapeutics — Hearing Loss Regeneration

[Gb3]

I hope they can unbind the results and find out those who faked their results and those that didn't. I think Phase 2a can be salvageable if they can get rid of the fake hearing tests by the participants and see if the results look better. A long shot is that Day 210 results may look even better and may show improvements like the positive anecdote had.

@Gb3, do you know how long it took for him to know when he felt like he started to notice a difference in his hearing during the 5 months after he took his last dose of FX-322? When did he start to feel like he had some improvements in certain hearing bands and word scores?

Also just to add 5 months is roughly 152 days so there may be a chance that 210 days may show some meaningful improvements and Phase 2a can be salvageable and allow them to move on to the pivotal phase early next year.

They tested his hearing after the injections. I think the improvement in hearing was during the 90 days but it was long before the tinnitus faded.

 

[dan]

I was afraid of this. Remember how we talked about the hearing aid industry. The audiologists wouldn't just let this happen.

 

[Aaron91]

I mean, what if FX-322 is killed/shut down (e.g., due to incapacity to objectively prove improvements in audiograms or EHF) before Frequency Therapeutics can conduct specific trials on specific subjective conditions? I feel like FX-322 is hanging by a thread, with Phase 1b serving as the thread. If they can't prove positive results from Phase 1b when the full readout is released in June (for what it is worth, I don't expect the full readout of Phase 2a to show anything meaningful), FX-322 and Frequency Therapeutics may well be doomed...

Thankfully they are well-funded to go to pivotal trial and get to market. But yes, if that fails too, they will need further funding to finance their other pre-clinical drugs. Not that it would matter to us anyway, because FX-322 would certainly be dead by then.

They tested his hearing after the injections. I think the improvement in hearing was during the 90 days but it was long before the tinnitus faded.

This is one thing that gives me a glimmer of hope for the 210-day readout. Hear me out.

Would patients consciously/subconsciously deflate their TFI scores as well as their WR scores? I find this unlikely as unlike the WR scores, Frequency Therapeutics didn't communicate a minimum TFI score as a selection criterion. Which would suggest then that if FX-322 works for tinnitus, the whole selection bias thesis falls apart if we don't see any improvements at the 210-day readout. When we pair this anecdotal experience of a patient improving after 90 days with some previous speculation that it could take the brain quite a few months to respond to the new input, I'm wondering whether it's possible we see statistically significant TFI scores in the full read-out. This would be huge, because tinnitus is an unmet meet and they would, presumably, make this a primary endpoint for their pivotal trial. It would be a major lifeline as far as confidence in the drug goes.

 

[FGG]

They tested his hearing after the injections. I think the improvement in hearing was during the 90 days but it was long before the tinnitus faded.

If this anecdote is correct, it means that some individuals did have audiogram changes but not at the group level (and apparently the more injections you got, the worse the outcome) but the problem is there were placebo improvements as well, which could actually be explained by people faking a worse audiogram to lower their PTA, which is extra muddying.

I hope the single dose severe group is more clear.

 

[aot]

 

[reggie green]

I wrote them an email asking what their next step would be.

 

[Aaron91]

This is an interesting thought. I think they didn't know about the Phase 2A that far back. Re-listening to the call this morning, I heard executives that were in the process of pivoting. If they had thought it out more, there would have been a much more controlled presentation with an actual deck, and some deep rationale for the new game plan, and how it would benefit investors + patients. None of that happened. They probably got the results, shit their pants, got back on the phone with the FDA, and came up with the single-dose strategy. This pivoting activity, and the way it was handled with a teleconference call, shows that they're in red-alert right now, and have been for a short period of time. Not to mention, the Q&A was happening in real-fucking-time... if they were in a prepared spin mode... there would have been no Q&A, it would have been polished

This pretty much summarises why I reacted so badly yesterday. The whole thing just stank from an investor POV, so many red flags. No visual data, no graphics, no presentation. It felt as if they were hiding behind the phone. But as I said later on yesterday after I cooled down, I too would be sheepish if I've just received data that's gone completely against what I was expecting with no individual level data to infer what the hell might have happened. The only thing that still bothers me is that they said they had been sitting on this for weeks. That tells me the data was so fucking bad there was no positive way of spinning anything. The group data must be damning, honestly. It's possible it looks even worse than we might imagine.

 

[100Hz]

This is an interesting thought. I think they didn't know about the Phase 2A that far back. Re-listening to the call this morning, I heard executives that were in the process of pivoting. If they had thought it out more, there would have been a much more controlled presentation with an actual deck, and some deep rationale for the new game plan, and how it would benefit investors + patients. None of that happened. They probably got the results, shit their pants, got back on the phone with the FDA, and came up with the single-dose strategy. This pivoting activity, and the way it was handled with a teleconference call, shows that they're in red-alert right now, and have been for a short period of time. Not to mention, the Q&A was happening in real-fucking-time... if they were in a prepared spin mode... there would have been no Q&A, it would have been polished.

So, I doubt they realized it back then, when setting up the new Phase 1bs. When they announced them, if you recall, they had to go back to many of their original backers to ask for another round of funding and dilute the overall holdings in the company. If the meeting had a negative spin at that time, I suspect some of the advisors would have backed out, and not been so willing to fork out millions for a pivot that early.

When Frequency Therapeutics announced the Age-Related Hearing Loss Study, it was a few months after new research was published (I believe by Mass Eye & Ear) that indicated Presbycusis was largely predicted by hair cell loss from observations of hundreds of cochleas. So, they probably looked at that and thought FX-322 would be a good treatment to expand the marketability of the product. Therefore, a Phase 1B would be needed to prove safety, and if successful would enable inclusion of that specific patient population into later studies.

For the Severe Hearing Loss Phase 1B trial. Total speculation, but they may have observed that there are a lot of progenitors left in a severe hearing loss cochlea. You might also notice that Severe was added to the open-label trial much later. So, it's hard to see that as a pivot.

Also, since the time of pivot, all the way through recently... those same backers continued to hold FREQ through the most recent drop. Why would they give more money to the firm, knowing that a significant loss was about to happen? Seems like dumb business from the backer's perspective.

Thanks @Diesel, great observation. They would have had so much time to prepare for that presentation if they had known. Plus the continued extra backing, it looks like confidence in FX-322 was there right up to the last minute and the extra trials were started with genuine belief in the product.

I learnt a lot yesterday though. The doubling of WR was always the focal point but I can't help thinking the EHF should have also been an important factor. I'm still going with a downgraded, cautiously optimistic stance that something IS being fixed that can't be measured (yet). The 1b results will need to be good.

 

[FGG]

This pretty much summarises why I reacted so badly yesterday. The whole thing just stank from an investor POV, so many red flags. No visual data, no graphics, no presentation. It felt as if they were hiding behind the phone. But as I said later on yesterday after I cooled down, I too would be sheepish if I've just received data that's gone completely against what I was expecting with no individual level data to infer what the hell might have happened. The only thing that still bothers me is that they said they had been sitting on this for weeks. That tells me the data was so fucking bad there was no positive way of spinning anything. The group data must be damning, honestly. It's possible it looks even worse than we might imagine.

"The placebo group dramatically improved their word scores, too" IS pretty bad data to receive. But weeks makes sense if you truly didn't know what the hell went wrong.

If Kevin Franck uncovered evidence that patients lied to get picked for the trial, I'm sure that took a bit of time to parse through especially without the individual unblinded data (you'd have to average what was in records with what was seen and/or go through the longer term records and see for instance that no one was naturally in the let's say 20 range, etc).

 

[athan86]

I wrote them an email asking what their next step would be.

Nice. When you receive a reply, kindly let us know.

 

[Zugzug]

Everyone's bitching that there's too many bulls in this thread, so I now have to become an analytical bear at times.

Here's what I've been thinking about with this placebo stuff. Based on their open-label Phase 1b trial (FX-322-111) as well as their analysis in the patent, it seems like they may have performed categorical tests as well.

The issue with the "faking" theory is that categorical tests significantly mitigate the effect of outliers. For example, Frequency Therapeutics said that the baseline audiograms of the placebos didn't match the screenings (presumably, there was a decent match for the treatment groups).

In all likelihood, let's be honest, it's not like the whole study consisted of fakers. We're probably talking about a few people roughing up the averages with outliers. But in the categorical tests, they are only looking for "responders", which based on trial FX-322-111, just requires >=10% absolute WR improvement, which isn't much at all.

In other words, it doesn't really matter if a faker in the placebo group jumps from 25% to 60%, they will count the same for this test as the person who jumped from 25% to 35%.

I can't imagine they didn't perform a test like this for "Responders (yes or no)". The fact that they didn't show it off is a bad sign. They have a defense for the more meaningful group average percentage improvements, if by chance, the outlying liars landed in the placebo group, but this still requires rigorous proof. They would have to statistically prove that the screenings don't match the baselines.

Anyways, I am concerned that they didn't really show off anything. They showed off the non-placebo-controlled open-label study. Not a good sign at all.

We may be saved in the two remaining Phase 1b studies are strong, but if they aren't, it's over for a long time. The company is not dead or broke, but something significant will have to change. This means delivery and possibly formulation. This would create a long delay.

 

[FGG]

Thanks @Diesel, great observation. They would have had so much time to prepare for that presentation if they had known. Plus the continued extra backing, it looks like confidence in FX-322 was there right up to the last minute and the extra trials were started with genuine belief in the product.

I learnt a lot yesterday though. The doubling of WR was always the focal point but I can't help thinking the EHF should have also been an important factor. I'm still going with a downgraded, cautiously optimistic stance that something IS being fixed that can't be measured (yet).

I come down to this binary: either some hearing got fixed or multiple Phase 1 patients lied dramatically about word scores (not just a little) for some unknown reason. I don't see any other choice. The placebo effect doesn't give you results that have historically been impossible.

 

[Guywithapug]

This pretty much summarises why I reacted so badly yesterday. The whole thing just stank from an investor POV, so many red flags. No visual data, no graphics, no presentation. It felt as if they were hiding behind the phone. But as I said later on yesterday after I cooled down, I too would be sheepish if I've just received data that's gone completely against what I was expecting with no individual level data to infer what the hell might have happened. The only thing that still bothers me is that they said they had been sitting on this for weeks. That tells me the data was so fucking bad there was no positive way of spinning anything. The group data must be damning, honestly. It's possible it looks even worse than we might imagine.

It certainly was bad news yesterday and I think most everyone was hacked off at the situation. It does appear that there is a lack of transparency and that what was provided was sugar coated by stating they would press ahead with single injections. In all honesty, I can understand their perspective though, they are committed to finding us a viable treatment whilst pioneering uncharted waters. They had to have damage control and keep some investor confidence.

I do want to thank you as well as @Zugzug, @Diesel and @FGG for explaining and theorizing things that some of us may not know here. May the future be with us.

 

[Aaron91]

Everyone's bitching that there's too many bulls in this thread, so I now have to become an analytical bear at times.

Here's what I've been thinking about with this placebo stuff. Based on their open-label Phase 1b trial (FX-322-111) as well as their analysis in the patent, it seems like they may have performed categorical tests as well.

The issue with the "faking" theory is that categorical tests significantly mitigate the effect of outliers. For example, Frequency Therapeutics said that the baseline audiograms of the placebos didn't match the screenings (presumably, there was a decent match for the treatment groups).

In all likelihood, let's be honest, it's not like the whole study consisted of fakers. We're probably talking about a few people roughing up the averages with outliers. But in the categorical tests, they are only looking for "responders", which based on trial FX-322-111, just requires >=10% absolute WR improvement, which isn't much at all.

In other words, it doesn't really matter if a faker in the placebo group jumps from 25% to 60%, they will count the same for this test as the person who jumped from 25% to 35%.

I can't imagine they didn't perform a test like this for "Responders (yes or no)". The fact that they didn't show it off is a bad sign. They have a defense for the more meaningful group average percentage improvements, if by chance, the outlying liars landed in the placebo group, but this still requires rigorous proof. They would have to statistically prove that the screenings don't match the baselines.

Anyways, I am concerned that they didn't really show off anything. They showed off the non-placebo-controlled open-label study. Not a good sign at all.

We may be saved in the two remaining Phase 1b studies are strong, but if they aren't, it's over for a long time. The company is not dead or broke, but something significant will have to change. This means delivery and possibly formulation. This would create a long delay.

@Zugzug I decided to go back just now and listen (with pain and difficulty) to the webcast. I don't think this has been mentioned yet, or at least it has not been appreciated, but apparently the actual improvement in WR scores in the treatment groups were not anywhere near what they had seen in the the original Phase 1/2 study as well as the recent off-label study. I had originally presumed that the treatment response in Phase 2 was the same as it was in Phase 1. It wasn't. This is why, Carl says, they make reference to this "dampening" effect with repeated doses that they don't quite understand on a molecular level.

Carl then goes on to say that the placebo group, in parallel, improved, but when they looked at their historical records compared to the baselines when they entered the trial, there were inconsistencies. He makes it clear - and I somehow missed this yesterday - that these inconsistencies were found across the WHOLE study. In other words, patients in BOTH the placebo and FX-322 group had inconsistencies in their historical records compared to their baseline values. It sounds like a catalogue of errors with an absolutely disastrous outcome. I'm starting to believe this might not be quite over.

Edit: When asked asked about whether they would change WR score entry requirements, Carl said they are looking forward to the opportunity to better define what a responder will look like, but he also said they felt ok with where they drew the line for WR scores. In other words, it sounds like they plan to dig through the individual data post 210-day readout and figure out who the idiots (non-responders/fakers) were.

 

[Philip83]

 

[Aaron91]

Chris Loose also seemed to answer a question I had yesterday (thanks @Tezcatlipoca for trying to help me with this) about whether we could glean anything from multiple ITT injections studies using steroids in the past. He said that ITT steroid injections are a "distinct example". In other words, we can't infer anything, I believe.

Edit: Carl also said that when you look at some of these ITT studies, the drop out rate is 10%, suggesting then that something uncomfortable is happening in the ear when you dose the cochlea repeatedly. I can't believe though they didn't look into this beforehand. I think in their attempt to chase those lower frequencies they got their fingers burnt big time.

 

[Chestnut94]

You only get one shot, do not miss your chance to blow.

 

[Jurger]

@Zugzug In other words, patients in BOTH the placebo and FX-322 group had inconsistencies in their historical records compared to their baseline values. It sounds like a catalogue of errors with an absolutely disastrous outcome. I'm starting to believe this might not be quite over.

This doesn't account for the fact that in their new Phase 1/2 study, which wasn't compromised:

1. There were no changes on the audiogram, not even in the extended high frequencies. Two studies have confirmed FX-322 reaches the cochlea.

2. Only 34% of patients responded with at least a 10% increase, with, granted, a small subset of patients doubling WR scores. That's underwhelming, to say the least. That's why they tried multiple dosing: the previous Phase 1/2 study indicated similar results.

They can't go to the FDA if it stays like this.

 

[andreimatei]

Everyone,

I'm a long time reader of this thread. I appreciate all of the wisdom found here - you guys are truly a great community.

Yesterday was awful for me. I couldn't get out of bed. My gut reaction to the perceived failure of FX-322 hit me hard. I've had so much hope hanging on this. I was also devastated at losing almost $40k and most of my stock account. (I am an idiot, wasn't diversified, not an experienced investor, and was emotionally invested, foolish I know.)

But today gives me some new hope. The hopeful and rational messages on the last several pages are encouraging and I believe we do have legitimate reasons to still have cautious faith. We aren't done yet with this story.

For those of you who are financially invested too and really down about it, just search Google for info on Oppenheimer's new price target of $20 and their reiterated "buy" rating as of late yesterday. Or B. Riley's revision to $35. Also, there are some good YouTube videos on price analysis and predictions too. It lowers my "freak out" levels just a bit. Might for you too.

I'm not going to post links here because I know it's not a financial board and I don't want to distract too much from the most important topic; efficacy. But just trying to spread some reasons for hope from the financial side if anyone needed that right now.

Anyway, even though it was a double gut punch yesterday, I think if we can stay calm and continue to be patient (it's so hard I know), we still have a chance for this to help us have better lives.

 

[Olive27]

I understand why but a lot of people on here are...

It is OVER for the foreseeable future.

 

[Zugzug]

@Zugzug I decided to go back just now and listen (with pain and difficulty) to the webcast. I don't think this has been mentioned yet, or at least it has not been appreciated, but apparently the actual improvement in WR scores in the treatment groups were not anywhere near what they had seen in the the original Phase 1/2 study as well as the recent off-label study. This is why, Carl says, they make reference to this "dampening" effect with repeated doses that they don't quite understand on a molecular level.

Carl then goes on to say that the placebo group, in parallel, improved, but when they looked at their historical records compared to the baselines when they entered the trial, there were inconsistencies. He makes it clear - and I somehow missed this yesterday - that these inconsistencies were found across the WHOLE study. In other words, patients in BOTH the placebo and FX-322 group had inconsistencies in their historical records compared to their baseline values. It sounds like a catalogue of errors with an absolutely disastrous outcome. I'm starting to believe this might not be quite over.

Edit: When asked asked about whether they would change WR score entry requirements, Carl said they are looking forward to the opportunity to better define what a responder will look like, but he also said they felt ok with where they drew the line for WR scores. In other words, it sounds like they plan to dig through the individual data post 210-day readout and figure out who the idiots (non-responders/fakers) were.

Keep in mind that I can't listen to the webcast so it creates an unfortunate lack of information for me.

Here's what I don't understand with the faker defense. Exactly as you say, it doesn't surprise me at all that they found inconsistencies in the whole study. Even just mathematically, let's say there are 10% fakers, or 9 people. There are 24 placebos and 72 treaters. We would expect about 7 fakers to go into the treater group and 2 fakers to go into the placebo group. That gives the treaters an advantage.

If we consider some extreme distribution of the fakers, so instead of a nice 7 and 2 split, let's say it's totally nutsy disproportionate. Like let's say 5 or more (of the 9) fakers landed in the placebo group, by chance. What is the probability of this?

It ends up being

(24 choose 5)*(72 choose 4)...
+ (24 choose 6)*(72 choose 3) ...
+ (24 choose 7)*(72 choose 2) ...
+ (24 choose 8)*(72 choose 1) ...
+ (24 choose 9)*(72 choose 0)
/
(96 choose 9),
which equals approximately 0.0406, or 4.06%.

In other words, in all likelihood, the "liar" factor should have been proportionately dispersed between groups.

Now, regarding the lying itself. Here's what I can't understand. So the theory is that lying got them in, but then their baselines were their real score (higher). In a sense, how does it matter how they got in if they still performed the study correctly?

If I understand what LeBel is really suggesting, it's that the treatment group was actually held back because the baselines started closer to the ceiling effect than they anticipated, disproportionately to the placebos.

In other words, I don't think the argument is that the placebo group got a bunch of liars. I think it's that across the board, their filter (which the bull thesis largely rested on) was nullified. Again, apologies if I am inferring incorrectly since I can't listen to it. So basically, the big mistake was that across, the board, there weren't enough true severe cases to enjoy a lack of ceiling effect.

This is fair, as the separation between placebos and treaters should be much more significant with less ceiling effect (see Phase 1).

It does seem like they are on to something with the "don't step on the lawn" analogy. However, if the Phase 1b remaining trials are not strong, this theory is gone.

Then what remains is really a defense of "everyone didn't have enough hearing loss," which is fair, but it's not great to still not see the treaters dominating the placebos.

 

[HootOwl]

After all this (and having read the various - and extremely informative - breakdowns from @Aaron91, @Diesel, @Zugzug and others) I'm still more or less Team Goop.

One question of mine is how this might affect the ability to apply for expanded use in the upcoming years. Due to the poor initial readout will demand drop? — since this was cited, at least initially, as the main reason for not making EU available.

Or will they circle the wagons and want to really fine tune the single dose injection before allowing anyone outside of a trial to try it?

I'd still kill to hit my high frequencies on the off chance it could reduce my electrical hissing. And others I'm sure feel the same.

 

[frpp]

I still feel dead after yesterday. What do we do now?

 

[RB2014]

I'm sorry about all of the stock losses everyone has posted. The insiders that have been selling should have been a warning sign to everyone. I understand the need to supplement a salary with stock sales, but to me the insider selling has just been too aggressive especially if they really thought this was going to take off.

A few people within Frequency Therapeutics are now rich beyond their imaginations and a lot of investors have been left holding the bag.

The results are disappointing.

 

[spacehunter]

I still feel dead after yesterday. What do we do now?

Jumping on the habituation bandwagon maybe?

 

[Zugzug]

Keep in mind that I can't listen to the webcast so it creates an unfortunate lack of information for me.

Here's what I don't understand with the faker defense. Exactly as you say, it doesn't surprise me at all that they found inconsistencies in the whole study. Even just mathematically, let's say there are 10% fakers, or 9 people. There are 24 placebos and 72 treaters. We would expect about 7 fakers to go into the treater group and 2 fakers to go into the placebo group. That gives the treaters an advantage.

If we consider some extreme distribution of the fakers, so instead of a nice 7 and 2 split, let's say it's totally nutsy disproportionate. Like let's say 5 or more (of the 9) fakers landed in the placebo group, by chance. What is the probability of this?

It ends up being

(24 choose 5)*(72 choose 4)...
+ (24 choose 6)*(72 choose 3) ...
+ (24 choose 7)*(72 choose 2) ...
+ (24 choose 8)*(72 choose 1) ...
+ (24 choose 9)*(72 choose 0)
/
(96 choose 9),
which equals approximately 0.0406, or 4.06%.

In other words, in all likelihood, the "liar" factor should have been proportionately dispersed between groups.

Now, regarding the lying itself. Here's what I can't understand. So the theory is that lying got them in, but then their baselines were their real score (higher). In a sense, how does it matter how they got in if they still performed the study correctly?

If I understand what LeBel is really suggesting, it's that the treatment group was actually held back because the baselines started closer to the ceiling effect than they anticipated, disproportionately to the placebos.

In other words, I don't think the argument is that the placebo group got a bunch of liars. I think it's that across the board, their filter (which the bull thesis largely rested on) was nullified. Again, apologies if I am inferring incorrectly since I can't listen to it. So basically, the big mistake was that across, the board, there weren't enough true severe cases to enjoy a lack of ceiling effect.

This is fair, as the separation between placebos and treaters should be much more significant with less ceiling effect (see Phase 1).

It does seem like they are on to something with the "don't step on the lawn" analogy. However, if the Phase 1b remaining trials are not strong, this theory is gone.

Then what remains is really a defense of "everyone didn't have enough hearing loss," which is fair, but it's not great to still not see the treaters dominating the placebos.

Let's ask ourselves this. If the unfairness worked in the opposite direction, would the company have had a problem?

So in my calculation above, let's say the 4.06% chance occurred and the placebo group had at least 5 fakers, while the combined treatment group had at most 4 fakers.

Then what would happen is the placebos would take the hit as the group with baselines closer to the ceiling effect (the bear thesis for Phase 1). This would help the drug pass the finish line. Would they point this out if it occurred?

The actual situation where it would help the placebos over the treaters is if a more extreme distribution of fakers was in the treatment group. So let's say the probability that at least 8 of the 9 fakers were in the treatment group; hence, the treatment group dealt with the ceiling effect at baseline more than the placebos.

The probability of at least 8 of the 9 fakers landing in the treatment group is

((24 choose 1)*(72 choose 8) + (24 choose 0)*(72 choose 9))/(96 choose 9) = 28.72%.

As we can see, there's a 71.28% chance (under the assumption of 9 fakers) that the "faker" effect was either proportional across groups or that it even helped the treatment group.

Is there something I'm missing? Obviously, the assumption of 9 fakers is completely speculative, but the ideas I'm proposing are similar with different numbers.

 

[Aaron91]

This doesn't account for the fact that in their new Phase 1/2 study, which wasn't compromised:

1. There were no changes on the audiogram, not even in the extended high frequencies. Two studies have confirmed FX-322 reaches the cochlea.

2. Only 34% of patients responded with at least a 10% increase, with, granted, a small subset of patients doubling WR scores. That's underwhelming, to say the least. That's why they tried multiple dosing: the previous Phase 1/2 study indicated similar results.

They can't go to the FDA if it stays like this.

I agree, and I had a hard time reconciling this yesterday as well. I think there are several possible explanations about the audiogram, depending on how far you're willing to stretch your imagination:

1) As others have said, the audiogram is an outdated diagnostic and highly specific. Is it possible that the audiogram does not test enough frequencies? Personally, I don't consider this to be a strong argument because surely over a large enough population you would still some kind of trend. I don't buy it, but it's possible.

2) The audiogram is not reflective of speech intelligibility. For what it's worth, my own audiogram is a case in point. I can tell you that as a life-long musician who's been trained to hear for things, my hearing is pretty terrible compared to what it used to be, but my audiogram looks pretty great.

3) The audiogram only reflects OHC performance and FX-322 focuses on IHCs (which are more important for speech intelligibility than OHCs) on its first pass. I had some reservations with this theory initially, but having done more reading since yesterday it seems very plausible. During development, you need an intact IHC in order for there to be a signal for the LGR5 cell to go to proliferate and divide into an OHC. This adds weight to the multiple-dosing mistake argument, in that each dose, regardless of whether it was FX-322 or placebo, wiped out any benefit from the previous dose. Keep in mind, I also shared a link yesterday that it takes 12 days for a hair cell colony to form. I have a feeling that if the doses were administered maybe a month or two apart, we would see exponential gains, assuming the regrown IHCs give the signal to the LGR5 progenitors to regrow into OHCs.

I agree that 34% of patients responding with a 10% increase doesn't look great in numbers, but we know that even a 10% improvement translates to a huge QoL improvement. It also says nothing about the patients' QoL who didn't have a 10% increase but still experienced, subjectively speaking, some measure of benefit.

All in all, it's very difficult for me to ignore that we have two separate, single-dose studies, showing benefit at a statistically significant level. Not to mention the fact that the second study could have been partially compromised to the same extent the Phase 2 study was. What I would like to know is: have Frequency Therapeutics gone and had a look at the historical records of the patients who responded in this off-label study and realised that their data was consistent in a way that the Phase 2 study was not?

Finally, on your point about the FDA, I strongly disagree. Frequency Therapeutics have Fast-Track approval. By default, this allows them to be in constant contact with the FDA. The FDA care about three things:

1) Safety
2) Meeting some kind of primary endpoint (whatever that is)
3) Identifying some kind of responsive population.

This is what we need to get the drug to market, which to me is the most important thing. We know the drug is safe. The audiogram doesn't need to be a primary endpoint as far as I understand it. In fact, Frequency Therapeutics don't even have to explain why the audiogram isn't responding while the WR scores are. If they can prove, on a statistically significant level, that a certain population is responding to a single dose in their WR scores without any adverse events, this drug gets approved to market. It's that simple. It's then up to us to get this off label and pay out of pocket if we have to.

 

[FGG]

Keep in mind that I can't listen to the webcast so it creates an unfortunate lack of information for me.

Here's what I don't understand with the faker defense. Exactly as you say, it doesn't surprise me at all that they found inconsistencies in the whole study. Even just mathematically, let's say there are 10% fakers, or 9 people. There are 24 placebos and 72 treaters. We would expect about 7 fakers to go into the treater group and 2 fakers to go into the placebo group. That gives the treaters an advantage.

If we consider some extreme distribution of the fakers, so instead of a nice 7 and 2 split, let's say it's totally nutsy disproportionate. Like let's say 5 or more (of the 9) fakers landed in the placebo group, by chance. What is the probability of this?

It ends up being

(24 choose 5)*(72 choose 4)...
+ (24 choose 6)*(72 choose 3) ...
+ (24 choose 7)*(72 choose 2) ...
+ (24 choose 8)*(72 choose 1) ...
+ (24 choose 9)*(72 choose 0)
/
(96 choose 9),
which equals approximately 0.0406, or 4.06%.

In other words, in all likelihood, the "liar" factor should have been proportionately dispersed between groups.

Now, regarding the lying itself. Here's what I can't understand. So the theory is that lying got them in, but then their baselines were their real score (higher). In a sense, how does it matter how they got in if they still performed the study correctly?

If I understand what LeBel is really suggesting, it's that the treatment group was actually held back because the baselines started closer to the ceiling effect than they anticipated, disproportionately to the placebos.

In other words, I don't think the argument is that the placebo group got a bunch of liars. I think it's that across the board, their filter (which the bull thesis largely rested on) was nullified. Again, apologies if I am inferring incorrectly since I can't listen to it. So basically, the big mistake was that across, the board, there weren't enough true severe cases to enjoy a lack of ceiling effect.

This is fair, as the separation between placebos and treaters should be much more significant with less ceiling effect (see Phase 1).

It does seem like they are on to something with the "don't step on the lawn" analogy. However, if the Phase 1b remaining trials are not strong, this theory is gone.

Then what remains is really a defense of "everyone didn't have enough hearing loss," which is fair, but it's not great to still not see the treaters dominating the placebos.

I actually have a theory about this and that is that people with truly bad word scores likely had audiograms that would put them more in line with the severe group (makes sense as they hear worse) and they didn't get any real borderline cases this time. This would mean most all or all of the low word scores were potentially fake but ironically the fake ones were preferentially selected to match the criteria.

The actual severe group shouldn't have this problem.

 

[Keith Handy]

After all this (and having read the various - and extremely informative - breakdowns from @Aaron91, @Diesel, @Zugzug and others) I'm still more or less Team Goop.

We need matching "Team Goop" t-shirts. Our mascot is an injured but determined-looking bull.

 

[Ilias T]

Man why so much negativity. They will keep working on it. From what I read, their multi dose shots did not work, and some additional unexpected results appeared.

The bias was also not stated as a fact, it was just suggested. Failures will happen, they did not give up on FX-322.