No, you're thinking of the word recognition (WR) score - that test is a "words in quiet" test. 4 patients saw statistically significant WR score improvements. I've attached the slide from their presentation which shows the difference between the two tests. A ~30% improvement was seen overall in the Words in Quiet test, while a ~20% improvement overall was seen in the Words in Noise test.
View attachment 43360
In my post I was referring to the Words in Noise test, which did not achieve statistical significance - but it is the subject of Frequency Therapeutics' tweet. Also of interest is that this type of hearing is supposedly improved by increasing synaptic connections (what OTO-413 does). Though Frequency Therapeutics may be trying to say that new hair cells help too.
Either way, my point stands. I'm on my phone right now, but assuming those slides are from their 19 January presentation, you will see on slide 61, if I recall correctly, that most patients did not have a measurable WR impairment. This would suggest then what I previously said, which is that the ceiling for improvement would be too low to be able to register a statistically significant improvement in the first place, especially given some, if not most of the patients only had mild hearing loss.
If I recall correctly, they were measuring a standard deviation in a x/50 WR test as 3 words. You need
at least more than one standard deviation (sometimes two or three or more depending on your test) to deem something as not being able to be explained by chance I.e. is statistically significant. With that in mind, let's assume for the sake of argument that 2 standard deviations (6 words) were required to observe statistically significant improvement. Given slide 61 says that most patients did not display measurable WR impairment, it wouldn't be far fetched to say many could have had a score of 45/50 or higher for a quiet WR test pre-trial. If they then scored 50/50 post FX-322, this 5 word improvement would still fall under 2 standard deviations. I'm personally not concerned, and I think this is why frequency did not focus on the mild hearing loss group for Phase 2, because they knew they'd have a better chance of demonstrating statistical significance in the moderate group, even if mild groups do indeed benefit.
Edit: when I said frequency were not focusing on mild groups, I didn't mean excluding. They are simply extending to moderately severe.
Edit 2: Ok, I realise I am still confusing the two tests, but it isn't obvious to me as to why one can't extend the same logic from quiet WR to word in noise. The ceiling may indeed be higher for word in noise in mild groups (it is for moderate/severe), although this is not immediately obvious to me, and the improvements observed may still be below the number of standard deviations required. This would also be consistent with the less strong p value, if i recall correctly, compared to the quiet WR test. As for possible explanations as to why, a mild hearing loss group may still have a low ceiling for improvement. Alternatively, the dosage schedule may have not been high enough to reach some critical frequencies. Phase 2 will answer this question. And of course, it's possible that given FX-322's hands are tied in regrowing synapses only where an OHC has lost its respective synapse, it's possible that a word in noise test is likely to see greater gains from a synaptopathy drug, such as OTO-413, as you have alluded to.
For the record I'm in bed on my phone about to sleep lol, so will come back on this again tomorrow with more clarity.