Algernon Pharmaceuticals Inc (AGNPF)

[medchem]

Trial design.

I posted the below comments on the Cannabix board. Given that FAIMS is being used by both companies I think it's relevant. What's perhaps most noteworthy is that Breathtec has a unique set of challenges. It's a bonus that the two hour time frame for detection is not an issue (the issue of metabolism of the active compounds that we look for is not an issue, in the sense that they aren't present in the body in small quantities for fletting periods of time, rather, there is tons of build up and they are their chronically, i.e. chronic diseases).

Also, patient populations with the disease, display the same biomarkers, there is less variablity to control for (i.e. less "confounding").

The sensitivity and specificity of the FAIMS device are the ultimate parameters that will determine the devices success in the market. The sensitivity and specificity for that matter, are the most important parameters of any medical investigation. The good news is, the sensitivity is VERY high on FAIMS given detection in parts per billion, and the abundant biomarkers are often present in parts per million... easily detectable.

However, the higher the sensitivity (i.e. being able to detect at all), the lower the specificity (i.e. being able to choose the RIGHT compound). Similar to my mention of signal to noise below. For Breathtec, the specificity is like choosing the correct mountain in the mountain range. It's actually not that bad.

I think we will get VERY VERY good news in the next week or so regarding FAIMS and next stage testing.

My guess at trial design/next step:

In order to prove proof of concept, the trial designer must attempt to test one variable at a time with as many of the other variables as possible being held stable. If they do not isolate each factor, then the trial would be biased by "confounding variables". In lay terms, one of the inherent difficulties is that there are a few factors at play. It's not as simple as alcohol detection. Here's why:

1. The marijuana: People who smoke marijuana most of the time never weigh out the exact quantity, so it's often hard to know how much someone has smoked. The marijuana can come in a variety of different strains, each has varying potency (i.e. different quantities of the psychoaactive metabolite THC-9 + a variety of other chemical compounds which can confound). These extra compounds often how similar molecular weight and are going to show up on FAIMS detection with the same mass/charge ratio and thus appear in the same detection window. Lastly, often drug dealers lace there marijuana, and so someone may be high, but unless FAIMS ALSO simiultaneously detects other substances (like cocaine), then the device would underestimate how much has been smoked. Thus, these factors need to all be considered.


2. The human who smokes it: Each and every person will metabolize the marijuana at a different rate. The time frame for clearance is already tough to capture in the real world, because THC-9 is so potent, someone may still be high, but if being detected at 2 hours time since smoking, the quantity in the body can still be SUPER low. So you need ultra high detection capability. With high detection however, comes an unwanted picking up of all the other small chemical enzymes and biomarkers in breath which are useless (unless your Breathtec). Fourier transformation and multi-scan approaches can improve signal to noise, picking up less crap. Similar to in nuclear magnetic resonance spectrometry.

Take Home Points: So the key here is to tune the device to look in the right mass to charge detection window at the right time, for THC-9 specifically, and to use FAIMS and fourier transformation to get the good signal to noise (thus finding only the one THC signal). This all hinges on the assumption that THC-9 is the psychoactive compound the person has smoked. It doesn't control how much THC-9 was in the particular strain, nor how fast the user metabolizes. THUS, a good trial design would take several people, have them all smoke one strain of weed, check at various time intervals, then have the same group smoke another strain, check at various time intervals. Then use the same strain, and use different people and check at various time intervals. In each case, you can control one variable and check the others, decreasing confounding.