Wonderful! Thank you. Scots is such a fascinating thing to listen to - just a beautiful illustration of how new languages develop

Thanks to @profmakris.bsky.social for invitation and editorial input and thank you to insightful, thoughtful reviewers who helped to improve the piece substantially. #andexanetalfa

7/ I conclude - some patients likely to benefit but we don't know who they are - probably those treated early. At the moment, lack of evidence for net clinical benefit makes me sceptical of all the options available.

6/ Thrombosis rates are high with andexanet alfa - statistically increased risk of thrombosis vs usual care. No mortality or disability benefit was shown (study underpowered for these) so we cannot know if there is net benefit.

5/ However, the lack of blinding & usual care control --> bias especially in context of surrogate endpoints - particularly the need for rescue therapy (which was part of the primary outcome). Need for rescue therapy was significantly better with andexanet in whole cohort analysis

4/ With a usual care control arm, blinding patients and physicians is impractical - given different methods of administration for PCC and andexanet, and the chaotic, time pressured situation. These trials are hard enough to run and recruit to.

3/ I argue that the usual care control arm is reasonable as evidence for PCC is scanty and is not licensed for FXa DOAC reversal.

2/ 530 patients, most with intracerebral haemorrhage randomised to andexanet or usual care in ANNEXA-I. Showed improved haemostatic efficacy with andexanet - a surrogate endpoint composed of haematoma expansion <35% at 12h, need for rescue therapy and change in NIHSS <7.

Maybe the best option is going to be B-cell delivered FVIII gene therapy. Excited to see how this goes.

They're planning on a pre-surgical trial against standard of care (which will be a messy control arm) so not going up against andexanet initially

Re antibodies, given it's a one off treatment for most, maybe not too worried although one would also worry about spreading of antibodies to endogenous FXa (although clearly unlikely and not seen with andexanet)

I think this is a promising agent but it's very different showing no thrombosis signals with d-dimer etc... than when you start giving it to real patients. The lack of effect on TFPI is reassuring though.

Thanks for reading. Check out Don't Just Read the Abstract podcast (on all major platforms). We have done a a two-part podcast on the ANNEXA-I trial.