ell I thought I'd be able to post in a day or so but there is a lot to unpack. So I'm breaking this up so I don't have to wait for days to post this!
Tonight's first part tackles ADP and a little on the Day 112 shift.
Part 2 will tackle more about the rest of the protocol changes and delve a bit more into what @Dungiven posted and his respective thoughts.
As always, my personal thoughts here, I could be wrong, always best not to rely on any one given poster and of course, DYOR
ADP WHO?
The most common pain scales used by far is WOMAC, KOOS is another one. What are the differences between these and what exactly is ADP anyway?
Womac stands for Western Ontario and McMaster Universities Osteoarthritis Index. See APPENDIX A for an example. The WOMAC scale consists of three subscales that add up to the overall scoring.1
KOOS is an extension of the WOMAC index.
ADP stands for Average Daily Pain and is a measurement of pain but instead of fewer and specified time points like Womac, it's conducted daily! It's then summarised at a weekly level.
The big difference here is Recall. What I mean by that is that with WOMAC, one has to think back to the day in question as to what they were feeling. A dear friend of mine recently said "Mozz, I can hardly think of what I did last Saturday, let alone what pain I was in!". In the case of ADP, it's done on a daily basis. No, the patient doesn't have to fill out an entire multi page questionnaire on a daily basis. It's just a short registering of a number on a pain scale. Easy. Done.
The big advantages2 of ADP is that it can be more accurate as the recall is on the spot. It can also tighten up the outliers. Thus, it can lead to a smoothing of the results. Here is an example utilising a drug called Zilretta.
An example of ADP measurements on a weekly basis, before any of you ask, Zilretta is a Corticosteroid. (Err... long term use aint great for your joints).
In a study utilising an intra-articular dose of Flexion's FX006 it was found that WOMAC seemed to produce better results when we are comparing Active -v- Active, but in the case of a study where it is Placebo -v- Active, ADP is better
"WOMAC was more sensitive and able to discern differences between active treatments".3
IS IT ADP OR BUST?
ADP is relatively new, there indeed have been a few trials that have used this measure but it still is relatively new.
Well here is the best news, PAR is going to be also doing WOMAC at the specified days as well. This was a relief for me as we know that iPPS works in a Bell Curve fashion. So there is a lovely peak effect there. Not just in terms of the amount of drug administered but indeed in terms of the efficacy profile of the drug over time. It typically builds up, plateaus to some extent and then drops away. We will check those Bell curves out in a sec.
FALLING DOWN THE CURVE?
The following has been embellished, not actual conversation:
So I asked a person suitably qualified to answer....waaaait a second, hold up.
We know the max peak of efficacy occurs on or around Day 56. You are now shifting it to Day 112. Isn't that a bit daft? (I actually suspected the answer but roll with me on this one). Well yes, that is a lower point on the efficacy curve for sure, this means more risk...BUT, we are comparing with 008....and to some degree 005. n was 15 and 56 in the Active respectively.
So I asked the expert, are we mitigating the risk by increasing the n? So we can now AFFORD to go out a number of additional days from 56 to 112, increase the n and have circa the same drug effect size?
The answer was a nod and a "Yes".
Most of you will get the meaning of the above, but I know a few may not fully pick it up, let me explain it.
We know drugs like iPPS typically work slowly, have a peak effect and then reduce over time. They follow a rough bell curve.
Here is a generic Bell curve:
But more importantly for us, here is an example of an actual curve from our results (005).
So I had to flip it and then I have superimposed green bell curves to make it a bit obvious to illustrate my point.
The original image is here:
You can see here that the peak is somewhere like Day 39 or 45 or so. But this is just two charts out of many, the peak would be something closer to Day 56 across a group of measures.
So this is the day we should always take, we want max bang for our buck, we want maximum chance of being right.
Right?
Hmmmmm ...not always, we will get to that soon later in this post.
What I asked my expert in the field is that we are now looking at a time point further to the right of that curve, so a LOWER efficacious point. This should normally send alarm bells ringing and raise some red flags. PAR has this WRONG....?!
Dear PAR, go back to Day 56, your peak effect day, but, and it is a big but, if you increase your n, you are effectively lifting that bell curve upwards.
How and why?
Because of the Percentage chance of success formula. Yeah that's a Mozz made up formula but let's now investigate further:
THE FORMULA
Dungiven mentioned something about Racing cars....
Dungiven is a smart cookie and he will know about Newton's 2nd law
F = m x a
Force = Mass x Acceleration
To get Verstappen's car to move fast as well as accelerate like no other...he needs Force to be high and Mass to be low....you get that, and you get increased ability to accelerate. That's how that formula above works. Increase the right hand side, decrease Mass and then Acceleration HAS to increase.
It's the same with our drug!
Huh?
In our trials we are governed by stats.
Oh it's all about stats.
A simplified version, we have three variables we can really control.
1) n
2) P
3) ES
What are those?
n are numbers of patients
P is confidence that we can prove the null hypothesis to be false (EG Patients are getting good results because the drug actually WORKS, NOT because of random chance)
ES is Effect Size of the Drug
So let's draw it up as a formula:
Well just before that, remember, this is a Mozz version of it, it's strictly for illustrative purposes only, the actual stats formula is way more complex and is quite different and much more involved. I'm using this example purely to give you an idea of what's happening.
ES = n x P on the assumption that ES (Drug Effect) is the same for our 008 and 012.
The best way for me to explain that is through some real examples.
Let's look at Aspirin. They had the following in their first trial:
CASE 1 - ASPIRIN
These dudes had such a big n, it's not funny.
I could make a Mozz Quiz © out of it but for brevity (and your sanity) I'll skip forward and just tell you the number.
22,000 Yeah that's a comma not a decimal point after the 22.
Their P was nuttily small:
p<.00001
This study was conducted over 5 whole years (And you complain our P3 going to be like 2 years with n = 466 ??!!)
They initially assumed that because their P value was somewhat (ok ok a LOT smaller) under 0.05 they had a successful study.
So as a result their Drug Effect size was also very small, but this got missed! The researchers were too focused on that very small P value.
Turns out they really did miss an important flag, because the drug effect size was small they missed dangers of people having an increased chance of Myocarditis!
"Focusing exclusively on the p value when the sample size is large can overstate the practical importance of one’s conclusions. In addition to a p value, researchers should report effect sizes and R2 so that readers can properly interpret the magnitude of their findings".4
They were forced to do re do the study! After that, they came out with warnings for prescribing Docs to watch out for people susceptible to heart failure and to avoid prescribing aspirin in these cases.
INTERESTING FACT - Aspirin's chief ingredient is obtained from the bark of a Willow tree. Pentosan is derived also from a bark, the bark of a European Beechwood tree.
CASE 2 PAR - 008
So let's do the same calc with 008.
n was teeeeeeny weeeny, 15 in the 2 x 2 dosing arm
P was pretty good, close to or better than 0.05 in a lot of endpoints and observables.
This means the only other variable left, HAS TO CHANGE...it meant that the drug effect size PAR was getting was HIGH
CASE 3 - PAR 012
This case study - one I really am going to love .
Now we plug in the values from CASE 2 into CASE 3 and you see the magic.
n is now HIGH (eg. it's a massive 233 -v- a paltry 15 that we had in 008)
Drug Effect stays the same as we are still using the same ingredients and the same dosing. ie ES is the same in 008 as it will be in 012 at Day 56
There is only ONE variable that can change....and that is P.
My friends, P goes DOWN to keep the same level of % of success if our n is higher than what it was in 008.
We already had fairly low P values anyway in 008, these are going to go much lower as n goes up. We aren't just adding in a few more n, we are increasing it by a factor of at least 15.
I believe it is this reason that PAR are getting confident about their P3.
IF IT AIN'T BROKE, WHY FIX IT?
Mozz, why change a non broken study (Racing car) when we know Day 56 is the sweet spot, the checkered flag of our trial, why the heck do we have to tinker with it?
The answer is a commercial one.
Ok we need a Mozz Story:
Imagine you go to your Doc with a sore knee thats getting harder to manage with them nasty NSAIDS...and you say, Doc, Enough. My pain is high, my function is low and on top of that I'm taking these Gut destroying NSAIDS, Not Monthly, Daily.
Enough!
So the Doc looks up the computer and this Zilosul® thing flashes up...he has a quick read of the label and enthusiastically says let's try you out on this one ...seems safe...seems durable...and some 'The Mozz' guy keeps jumping up and down about it too...
So the Doc looks up the dosing regime and explains it to you.
You say no way, not taking it.
Huh?
Why not!?
Well the patient says...mate, you want me to poke myself 12 times over 6 weeks and at the end of it I get efficacy to Day 56 which is err...like...*calculating*....56 - (6 weeks x 7 days = 42) 42 days which is 14 days??
So I go through the hassle of coming to you that many times or self injecting for a pathetic measly 14 days of relief? Mate, corticosteroids gives me a blooming 3 months of pain relief !!
Yes well now 112 day shift from day 56 seems much better and much more competitive, much more commercial.
To say nothing of the damage that corticosteroids can do over a few years to the very thing you are trying to protect!
So in other words we now are comparing apples with apples in terms of drug effect over a similar duration....Corticosteroids gives around 3 months of relief...112 days is about (112 -42) 72 days about 2.5 months. Of course we should see some efficacy out to day 404 (365 after last dose)...in which case this really will be highly commercial.
At the end of the day we want a great chance of success in our last trial BUT we also want and need, to to be commercial.
The market will LIKE a drug that works well and is safe for a couple weeks according to the label...
The Market will LOVE and DEMAND a drug that does the above but lasts at least as long as Corticosteroids and possibly longer in a fair few cases.(subject to Day 404 read).
There definitely is some correlation between the VAS pain scale and WOMAC but this seems to be particularly prevalent when comparing weight bearing activities (eg stair climbing).
The other possible variability between the two measures can occur with patients that have strong but intermittent pain versus those that have that more constant pain, as observed in one such study:
"...reliability of retrospective measurements may be lower in patients with a higher intensity of intermittent pain."
...indicating that there are indeed some merits of ADP.
ADP CONCLUSION
If Par were simply introducing a new method to document pain out of the blue and introducing this into a Phase three when, as Dungiven and a few others also infer, we have no history of this, I would be a little nervous. However, I believe it is more the FDA that suggested it. I also am heavily relieved that we still have WOMAC. We have some very strong evidence of WOMAC at 56 days and even after that, if you go back through the charts that Dungiven has recently reproduced.
I am also very much at ease when I know n is significantly increased.What does n have to do with pushing our interim read from Day 56 to Day 112?Well we know that an increased n, given Drug Effect Size is constant, pushes DOWN the value of P and this will be true even if time is shifted from Day 56 to day 112.. Get a low P and we could actually be allowed to sell this drug into major markets a lot sooner than we all surmise.
All very excellent Mozz but how does it help me and my shares?
Well that's easy. We are now looking at not one but two pain indicators...WOMAC and ADP. ADP gives us some advantages over WOMAC in that we get easy recall from the patient, patient convenience in that they don't have pages of info to fill out and we get smoothing of data.
It's also my belief that it wasn't PAR that requested this, it was the FDA that suggested it (Personal opinion, unsubstantiated comment).
Finally the winner of this commercial race isn't just the one that can go the fastest or the one that drives best only at the Peak of the curve race. It's the one that can make a commercial plan that's robust, that adheres to what the racing officials want and that's absolutely strategic.
- Mozz
Personal interpretations and views stated, not advice
APPENDIX A
EXAMPLE of WOMAC
(20min delay)