ImetelChat

Can someone help me better understand the "statistically powered" comments on Slide 17 (shown below) of latest corporate presentation? I'm interested to know how to properly interpret them, and how a couple of trial scenarios would be viewed relative to the trial design. For example, what would it mean if the placebo rate exceeds 7.5%, the imetelstat arm comes in just under 30%, or both happen simultaneously?

Statistically well-powered trial:
• Designed with >85% power to detect statistically significant
difference in 8-wk TI rate between placebo and imetelstat
(one-sided alpha=0.025)
• Conservative powering assumptions: an 8-wk TI rate of
7.5% in the placebo arm vs. 30% in the imetelstat arm

HI, we really need a statistician to properly address your question but I will give you my kindergarden interpretation. I see this trial design as a direct shot accross the bow of the Medalist study. Intentionally. Dr. Scarlett and company knew this would be Imetelstat's primary competition and wanted to differentiate as much as possible. The Medalist trial had a spontaneous 8 wk TI rate in the untreated placebo arm of 13 %. Pretty high. 38% achieved TI at 8 weeks in the treatment group. Also this study was potentially skewed to favor the low and very low risk MDS groups (82%). I think Geron is saying they are allowing for a higher placebo arm bar (more stringent criterion) by expecting a significantly lower spontaneous TI rate in the placebo arm, this could make for quite the p value. Also I could not find what the statistical power for MEDALIST was but in general, statistical power > than 80% is strong and therefore 85% even better. They difinitely are hoping to show strong differentiation and superiority of Imetelstat by setting the bar higher in this way. Other explainations or revisions welcome. bp

Do I understand the 13% spontaneous TI in the placebo arm to mean that 13% of the patience in that group achieved transfusion independence on just the placebo? What does "spontaneous" mean in this instance? That they just, over the course of so many weeks, while taking basically a sugar pill, stopped needing transfusions?
Thanks for any reply.
G.

Yes, that’s what it means. Presumably a similar number would have achieved TI in the treatment group had they not received treatment. However, treatment resulted in a highly statistically significant change in the treatment group compared to the placebo group that was not due to chance. By requiring a more stringent transfusion requirement for entry into either arm, Geron feels the likelihood of spontaneous achievement of TI will be much lower and the proof of med much stronger.

The two metrics are likely to move together in same direction. A “healthier” average patient population is likely to yield higher 8-wk TI rates in both arms while a sicker patient population (ie higher average transfusion burden at enrollment) is likely to yield lower 8-wk TI rates in both the placebo and the imetelstat arms. Thus, I wouldn’t expect to see an outcome with a relatively high placebo rate and a relatively low response for patients in the imetelstat arm.

Exactly. When Dr Scarlett talks about a highly differentiated profile I think this can be taken in multiple levels. One is different MOA, NME and also significantly better than the competition with a study design set up to demonstrate definitive differentiated superiority.

If memory serves, I believe an 8-wk TI period counts regardless of timing. Stated differently, a healthy patient who achieves an 8-wk TI period early in their treatment period counts the same as a sicker patient who realizes a 8-wk TI period later in their treatment period.

Further, I recall from prior studies that the patients who respond tend to respond within a 3-4 month time period. Thus, a patient who’s treatment regimen was disrupted by COVID upon entering the trial, may still have had adequate time to respond once they were able to get their infusions based on the trial protocol.

Correct. 8 consecutive weeks anytime in the study period.