Volume reduction with surgery for ICH: when is it effective? Analysis of the CT scans from the STICH II trial.

Background: The surgical trial of lobar intracerebral haemorrhage (STICH II) was a randomised controlled trial evaluating early surgical removal of a clot. This paper investigates volume change in both arms of the trial with respect to Extended Glasgow Outcome Scale (GOSE) groups.

Methods: Patients randomised into STICH II had an initial diagnostic CT and a second CT 5 days after randomisation. Each scan was anonymously assessed by at least two central readers. An analysis of agreement between the two readers was conducted using kappa tests and intraclass correlation. The change in volume in both the early surgery (ES) and the initial conservative treatment (ICT) arms were analysed with respect to the six-month GOSE outcome.

Results: Of the 597 patients randomised in the trial there were 582 pre-randomisation scans and 566 5-day scans available for analysis of agreement. There was good agreement between the assessors for the radiological inclusion criteria including volume (ICC = 0.87) and this was better than the agreement between the assessor and local investigator (ICC = 0.73). There were 526 patients with two scans available for analysis of change in volume measurement. The median percentage change in volume for the ES group was a reduction of 92.4% (IQR 75.6%, 99.0%) while for the ICT group, in which some cases crossed over to delayed surgery, it was only 5.7% (IQR 16.4% increase, 29.5% reduction). ES patients with almost complete removal (99-100%) had the best outcome with only 30% dead or lower severely disabled. For the ICT group outcome was related to the final volume: the smaller the final volume the better the outcome.

Conclusions: This analysis provides evidence for central assessments of scans in exploratory analyses and further information regarding the potential advantage of early and more complete clot removal on outcome in ICH and should inform the planning of future trials. ISRCTN22153967.