Between June, 2005, and March, 2007, 186 patients from 52 hospitals in 12 countries in Europe, Asia, Australia, and North America were randomly assigned to either 90 µg/kg desmoteplase (n=60), 125 µg/kg desmoteplase (n=68), or placebo (n=65). shows the trial profile. Seven patients were randomised but not treated and they were not included in the analysis. 158 patients completed the study.
The study groups were well balanced with regard to baseline characteristics (). The patients had an overall median age of 71·0–73·5 years, a stroke of mild to moderate severity (median baseline NIHSS score of 9 points), and most (63% [117 of 186]) were treated within 6–9 h of the onset of the symptoms of stroke. 66% (122 of 186) of patients were randomised on the basis of MRI screening. 30% (53 of 179) had vessel occlusion (TIMI 0–1) at randomisation: 20% (12 of 61) on CT angiography and 35% (41 of 118) on MR angiography. The absolute core lesion volumes (overall median volume 10·6 cm3) and mismatch volumes (overall median volume 52·5 cm3) were small across the study groups. The relative mismatch volumes were high across the study groups, with an overall median of 530% (IQR 200% to 1700%).
Characteristics of patients at baseline
There was no statistically significant difference in clinical outcome between the three groups. shows the composite responder rate (clinical improvement) at day 90, as defined by the primary endpoint.
Primary effi cacy endpoint (composite responder rate at day 90) in the intention-to-treat population
The proportion of patients with serious adverse events was 35% (20 of 57) for 90 µg/kg desmoteplase, 36% (24 of 66) for 125 µg/kg desmoteplase, and 29% (18 of 63) for placebo. The most common serious adverse events were ischaemic stroke (11% [20 of 186]), pneumonia (7% [12 of 186]), decreased haemoglobin (6% [11 of 186]), and intracranial haemorrhage (5% [10 of 186]).
Symptomatic intracranial haemorrhage within 72 h after treatment was observed in 4% (5 of 123) of patients in the desmoteplase groups—3·5% (2 of 57) in the 90 µg/kg group and 4·5% (3 of 66) in the 125 µg/kg group—and none of the placebo-treated patients. One of the two symptomatic intracranial haemorrhages in the 90 µg/kg desmoteplase group was due to worsening of a baseline haemorrhagic infarction, which constituted a serious protocol violation. The observations of symptomatic intracranial haemorrhage made by principal investigators matched the central blinded assessment except in one case, which was deemed as indeterminate (it met the protocol definitions of neither symptomatic nor asymptomatic) according to central assessment.
Asymptomatic intracranial haemorrhage was seen within 72 h of giving study medication in 35% (20 of 57) of patients treated with 90 µg/kg desmoteplase, 33% (22 of 66) of patients treated with 125 µg/kg desmoteplase, and 33% (21 of 63) of placebo-treated patients.
The application of the most conservative definition of the National Institute of Neurological Disorders and Stroke (NINDS) study criteria, in which acute neurological worsening of 1 point or more in the NIHSS is deemed relevant, the rates of symptomatic intracranial haemorrhage within 72 h of giving study medication were ·8% (5 of 57) for 90 µg/kg desmoteplase, 10·6% (7 of 66) for 125 µg/kg desmoteplase, and 14·3% (9 of 63) for placebo. By use of the ECASS II definition, in which bleeds that occur up to 7 days and are associated with a worsening of the NIHSS of 4 points are counted, the rate of symptomatic intracranial haemorrhage was 5·3% (3 of 57) for 90 µg/kg desmoteplase, 9·1% (6 of 66) for 125 µg/kg desmoteplase, and 4·8% (3 of 63) for placebo.
13 major haemorrhagic events were reported by 5% of patients (3 of 57) in the 90 µg/kg desmoteplase group, 8% (5 of 66) in the 125 µg/kg desmoteplase group, and 6% (4 of 63) in the placebo group. Three events, one in each study group, occurred within 24 h of study drug administration; 10 events occurred later than 72 h post-treatment. 21 deaths occurred within 90 days (11% overall mortality rate). Four patients died in the placebo group, three in the 90 µg/kg desmoteplase group, and 14 in the 125 µg/kg desmoteplase group. No anaphylactic reactions occurred.
In October, 2006, after 170 patients had been recruited, the data monitoring committee temporarily halted enrolment in the trial to investigate the increased frequency of deaths in the 125 µg/kg desmoteplase group. After a review of the unblinded data, the study was restarted on the recommendation of the data monitoring committee. Neither the details of the data monitoring committee’s unblinded review nor its rationale for halting the trial were discussed with the other study participants until after the database was locked.
A blinded adjudication of the causes of deaths in DIAS-2, done by members of the steering committee and the data monitoring committee after termination of the study, attributed four of the 14 deaths in the 125 µg/kg desmoteplase group to the index stroke (including three symptomatic intracranial haemorrhages), four to recurrent stroke and its complications, and the rest to miscellaneous reasons (). Three deaths, all due to symptomatic intracranial haemorrhage, were assessed as causally related to desmoteplase and one death due to recurrent stroke was deemed as unlikely to be related. Thus, 11 of 14 deaths were adjudicated as death due to stroke or reasons unrelated to the study drug, with most being late deaths (). In general, patients who died in the study were older than those who survived (median age 79 years vs 72 years), had higher baseline NIHSS scores (median score 15 points vs 9 points), and had larger core lesion volumes on MRI (median volume 25·8 cm3 vs 8·6 cm3).
Kaplan–Meier estimates of survival
Comparison of baseline NIHSS scores, mismatch volumes, core lesion volumes, vessel occlusion, and age between the placebo and desmoteplase groups within each imaging subgroup (MRI and CT) showed no statistically significant differences (). A comparison of the MRI and CT subgroups, however, showed a statistically significant difference for absolute and relative tissue at risk volume (higher in MRI; p<0·0001 and p=0·0001, respectively) and NIHSS score (higher in CT; p=0·007). Although low in both groups, the rate of vessel occlusion at baseline in patients who were assessed with CT was significantly lower than in the patients assessed with MRI (20% [12 of 61] vs 35% [41 of 118]; p=0·04). However, vessel occlusion at baseline was associated with higher NIHSS scores when assessed with CT than with MRI (median score 18·5 points vs 10·0 points).
Selected baseline characteristics that potentially predict a clinical response by imaging modality (MRI vs CT)
The central assessment of baseline scans for randomised patients confirmed a qualifying mismatch in 88% (56 of 64) of CT-screened patients and in 84% (103 of 122) of MRI-screened patients. The non-confirmed cases had matched core and hypoperfusion (n=19), no perfusion abnormality (n=4), or were technically inadequate to assess (n=4).
There was no significant effect of treatment on core lesion volume. Patients treated with placebo showed a small decrease in core lesion volume between baseline and day 30 (median change –10·0% [–0·9 cm3]), whereas patients treated with 90 µg/kg desmoteplase had an increase in core lesion volume of 14% (0·5 cm3) and patients treated with 125 µg/kg desmoteplase had a 10·8% (0·3 cm3) increase in core lesion volume. The patients who had the higher dose showed an inconsistent pattern, with an increase in core lesion volume of 43% (2·9 cm3) in patients screened with MRI compared with a decrease of 44·7% (4·2 cm3) in the patients screened with CT. The percentage changes have to be regarded with caution because of the small initial volumes. Irrespective of treatment assignment, patients with good clinical outcome were more likely to have no lesion growth, defined as any increase in lesion size (63% [46 of 73], 95% CI 51%–74%), whereas those with poor clinical outcome were more likely to have lesion growth (62% [53 of 86], 95% CI 51%–72%). The measurement of core lesion volume by the central image reading group showed greater inter-rater variability for CT than for MRI (inter-rater difference of >30% and >5 cm3: 41% [26 of 64] for CT vs 10% [12 of 122] for MRI).
The clinical response rates in patients who qualified for the study on the basis of MRI were compared with those who qualified on the basis of CT. We found no statistically significant overall difference in response rates: 90 µg/kg desmoteplase MRI 54% [21 of 39] vs CT 33% [6 of 18]); 125 µg/kg desmoteplase MRI 38% [17 of 45] vs CT 33% [7 of 21]); placebo MRI 45% [17 of 38] vs CT 48% [12 of 25]. The difference between MRI and CT response rates for the lower dose can be explained by an imbalance in baseline NIHSS scores (median 8 points [MRI] vs 12 points [CT]).
We also studied the effects of time from stroke onset to treatment and of sex. The clinical response rates in patients who were treated 3–6 h after stroke onset compared with those treated after 6–9 h were: 90 µg/kg desmoteplase 42% (8 of 19) versus 50% (19 of 38); 125 µg/kg desmoteplase 46% (11 of 24) versus 31% (13 of 42); and placebo 50% (13 of 26) versus 43% (16 of 37). Women generally had a worse outcome than men (90 µg/kg desmoteplase 43% [13 of 30] vs 52% [14 of 27]; 125 µg/kg desmoteplase 32% [12 of 37] vs 41% [12 of 29]; and placebo 38% [10 of 26] vs 51% [19 of 37]).
The clinical response rates in patients with vessel occlusion (TIMI 0–1) were assessed retrospectively: 90 µg/kg desmoteplase 36% (5 of 14); 125 µg/kg desmoteplase 27% (6 of 22); and placebo 18% (3 of 17). The corresponding rates in patients with no occlusion (TIMI 2–3 points) were: 90 µg/kg desmoteplase 50% (20 of 40); 125 µg/kg desmoteplase 40% (16 of 40); and placebo 57% (26 of 46).