Statistical significance is considered p =< .05. p for halt was much likely much lower due to a broad confidence interval for overall response rate because of the lower number of patients. The p =< .005 mentioned here sounds about right.

The much lower p for halt is basically a way for the DSMC to cover their and the company's asses to make absolutely sure if there's a halt there will be an approval. But the p =< .005 is what you'd see in a BLA "is proven effective" submittal not an EUA "may be effective". With 193 patients in a "may be effective" scenario p =< .05 should be sufficient to prove that out.