DESTINY II
Hemicraniectomy in Older Patients with Extensive Middle-Cerebral-Artery Stroke
Year of Publication: 2014
Authors: Jüttler E, Unterberg A, Woitzik J, et al; DESTINY II Investigators
Journal: N Engl J Med
Citation: N Engl J Med 2014;370(12):1091-1100. DOI: 10.1056/NEJMoa1311367
PDF: Clinical Question
In patients ≥61 years of age with malignant middle-cerebral-artery infarction, does early decompressive hemicraniectomy improve survival without severe disability (mRS 0-4) at 6 months compared with conservative ICU treatment alone?
Bottom Line
In patients ≥61 years with malignant MCA infarction, early hemicraniectomy roughly doubled the 6-month rate of survival without severe disability (38% vs 18%; OR 2.91, P=0.04), driven almost entirely by a large mortality reduction (33% vs 70% at 6 mo). However, no survivor in either arm reached mRS 0-2 and the majority of surgical survivors had mRS 4-5 (dependent for most bodily needs) — an important, values-sensitive tradeoff for shared decision-making.
Major Points
- Investigator-initiated, multicenter (13 German sites), prospective, randomized, open, 1:1 trial with blinded end-point assessment; sequential design using Whitehead triangular test (PEST v4.4) with prespecified interim analyses.
- 112 patients aged 61-82 years (median 70) randomized within 48 h of MCA stroke onset (Aug 2009 - March 2012); trial stopped for efficacy after DSMB review; ITT arms 49 (hemicraniectomy) vs 63 (control).
- Primary endpoint mRS 0-4 at 6 months (survival without severe disability): 20/49 vs 10/63 (bias-corrected 38% vs 18%; OR 2.91, 95% CI 1.06-7.49, P=0.04); raw mRS proportional-odds OR 3.97 (95% CI 1.39-8.76, P=0.01).
- 6-month mRS distribution: 0-2 = 0% in both arms; 3 = 7% vs 3%; 4 = 32% vs 15%; 5 = 28% vs 13%; 6 (dead) = 33% vs 70%.
- 12-month survival: hemicraniectomy 57% (95% CI 42-72; 27/47) vs control 24% (95% CI 14-37; 15/62). At 12 months, mRS 5 rate was 19% surgery vs 8% control among all randomized.
- 12-month secondary outcomes in ITT with worst-case imputation for deaths (NIHSS, Barthel, SF-36, HDRS, EQ-5D) all favored hemicraniectomy (P≤0.002); among survivors, no statistically significant between-arm differences on any secondary scale.
- Cause of death: 34/47 (72%) of control deaths were early neurologic herniation vs 4/20 (20%) in surgery arm; surgery deaths were more often delayed (pneumonia/sepsis, PE, contralateral infarct, unknown).
- Safety: 88 vs 84 SAEs; more infections with surgery (attributed to longer ICU stay); 23 hemicraniectomy-related complications (5 hemorrhages, 10 pain, 1 hygroma, 1 anesthesia event); no excess non-neurologic mortality with surgery.
- Retrospective consent to treatment: 63% surgery vs 53% control among survivors able to answer (25/42 survivors could not answer due to severe aphasia/cognitive impairment).
- Quality-of-life scores (EQ-5D, SF-36) among survivors were markedly reduced vs older stroke and general-population norms but similar to survivors of severe TBI, SAH, or ICH — comparable to younger malignant-MCA survivors.
Design
Study Type: Randomized, controlled, open-label, multicenter, sequential trial with blinded outcome assessment
Randomization: 1
Blinding: Open-label treatment; outcome assessment by physicians not otherwise involved in trial or patient care (PROBE design)
Enrollment Period: August 2009 - March 2012 (13 sites, all in Germany; recruitment stopped early by DSMB)
Follow-up Duration: 6 months (primary endpoint) and 12 months (secondary endpoints)
Centers: 13
Countries: Germany
Sample Size: 112
Power Calculation: 90% power to detect a difference between assumed success rates of 31.0% (hemicraniectomy) vs 8.6% (control) — corresponding to a log odds ratio of 1.56 — using a two-sided alpha of 5% with sequential Whitehead triangular test.
Analysis: Intention-to-treat; bias-corrected odds ratio from sequential analysis (Whitehead); per-protocol and sensitivity analyses (dichotomized mRS 0-3 vs 4-6; raw mRS scores) in supplement; worst-case imputation for missing endpoints in patients who died and last-observation-carried-forward for missing mRS/NIHSS in survivors.
Inclusion Criteria
- Age ≥61 years
- Clinical acute unilateral middle-cerebral-artery infarction
- Symptom onset <48 hours before initiation of treatment
- NIHSS >14 (non-dominant hemisphere) or >19 (dominant hemisphere) WITH reduced level of consciousness
- Ischemic infarction of at least two-thirds of the middle-cerebral-artery territory, including the basal ganglia, on brain imaging
- Written informed consent from patient or legally authorized representative
Exclusion Criteria
- Preexisting modified Rankin scale score >1
- Preexisting Barthel index <95
- Absent pupillary reflexes
- Glasgow Coma Scale score <6
- Intracerebral hemorrhage or other associated brain lesions
- Contraindications to surgery
- Estimated life expectancy <3 years
Baseline Characteristics
| Characteristic | Hemicraniectomy (n=49) | Control (n=63) |
|---|---|---|
| Median age, y (range) | 70 (62-82) | 70 (61-80) |
| Male, n (%) | 25 (51) | 31 (49) |
| Female, n (%) | 24 (49) | 32 (51) |
| Preexisting mRS 0, n (%) | 39 (80) | 53 (84) |
| Preexisting mRS 1, n (%) | 10 (20) | 10 (16) |
| Preexisting Barthel median (range) | 100 (95-100) | 100 (95-100) |
| MCA territory only, n (%) | 36 (73) | 40 (63) |
| MCA + ACA, n (%) | 11 (22) | 18 (29) |
| MCA + PCA, n (%) | 2 (4) | 5 (8) |
| Dominant-hemisphere stroke, n (%) | 16 (33) | 25 (40) |
| GCS median (range) | 12 (6-15) | 10 (6-15) |
| NIHSS assessable, n (%) | 34 (69) | 39 (62) |
| NIHSS median (range) | 20 (15-40) | 21 (15-38) |
| Onset-to-randomization median, h (range) | 25 (12-49) | 26 (9-47) |
| Onset-to-hemicraniectomy median, h (range) | 28 (16-50) | |
| Adherence to assigned treatment, n (%) | 48 (98) | 62 (98) |
| Onset-to-hemicraniectomy | NA |
Arms
| Field | Hemicraniectomy | Control |
|---|---|---|
| Intervention | Early decompressive hemicraniectomy (bone-flap diameter ≥12 cm) with duroplasty within 48 h of stroke onset and not later than 6 h after randomization, plus ICU care. | ICU stroke care per consensus protocol: osmotherapy (mannitol, glycerol, or hypertonic hydroxyethyl starch), sedation, intubation and mechanical ventilation, hyperventilation, and buffer solutions. |
| N | 49 | 63 |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Survival without severe disability, defined as mRS 0-4 at 6 months (intention-to-treat, bias-corrected sequential-design estimate). | Primary | 10/63 (bias-corrected 18%) | 20/49 (bias-corrected 38%) | 5 | 0.04 |
| mRS 0-2 at 12 months (no or mild disability) | Secondary | 0/62 (0%) | 0/47 (0%) | ||
| mRS 3 at 12 months (moderate disability) | Secondary | 2/62 (3%) | 3/47 (6%) | ||
| mRS 4 at 12 months (moderately severe disability) | Secondary | 7/62 (11%) | 15/47 (32%) | ||
| mRS 5 at 12 months (severe disability) | Secondary | 5/62 (8%) | 9/47 (19%) | ||
| mRS 6 at 12 months (death) | Secondary | 47/62 (76%) | 20/47 (43%) | <0.001 (ITT worst-case); 0.73 among survivors | |
| 12-month survival (Kaplan-Meier) | Secondary | 24% (95% CI 14-37); 15/62 | 57% (95% CI 42-72); 27/47 | ||
| Barthel index 60-100 at 12 months | Secondary | 8/13 (62%) | 24/27 (89%) | 0.002 (ITT worst-case); 0.34 among survivors | |
| NIHSS 8-16 at 12 months (moderate impairment) | Secondary | 4/10 (40%) | 14/22 (64%) | <0.001 (ITT); 0.70 among survivors | |
| SF-36 mental component 51-100 at 12 months | Secondary | 6/12 (50%) | 10/25 (40%) | <0.001 (ITT); 0.86 among survivors | |
| HDRS 0-19 at 12 months (no severe depression) | Secondary | 5/6 (83%) | 18/18 (100%) | <0.001 (ITT); 0.97 among survivors | |
| EQ-5D visual-analogue score 51-100 at 12 months | Secondary | 2/10 (20%) | 6/22 (27%) | <0.001 (ITT) | |
| Retrospective consent to received treatment (among survivors able to answer) | Secondary | ≈53% | ≈63% | ||
| Total serious adverse events (count) | Safety | 84 events | 88 events | ||
| 6-month mortality | Safety | 70% | 33% | ||
| 12-month deaths (total) | Safety | 47/63 (75%) | 20/49 (41%) | ||
| Early (0-14 d) death from herniation | Safety | 34/47 of deaths (72%) | 4/20 of deaths (20%) | ||
| Early (0-14 d) non-neurologic death (pneumonia/MI/sepsis) | Safety | 2/47 of deaths (4%) | 1/20 of deaths (5%) | ||
| Late (15 d-12 mo) death - pulmonary embolism | Safety | 0/47 | 2/20 of deaths (10%) | ||
| Late (15 d-12 mo) death - pneumonia/sepsis | Safety | 4/47 of deaths (9%) | 5/20 of deaths (25%) | ||
| Late (15 d-12 mo) death - new contralateral infarct | Safety | 2/47 of deaths (4%) | 2/20 of deaths (10%) | ||
| Late (15 d-12 mo) death - myocardial infarction | Safety | 2/47 of deaths (4%) | 0/20 | ||
| Hemicraniectomy/bone-flap complications (total 23 events) | Safety | NA | 5 hemorrhages, 10 cases of pain requiring pharmacologic treatment, 1 hygroma, 1 anesthesia-related event, 6 nonspecified (5 classified as SAEs) | ||
| Infections (qualitative) | Safety | Fewer infections | Numerically more frequent in hemicraniectomy group (attributed to longer ICU stay and longer survival) |
Subgroup Analysis
No formal subgroup effect modification reported; the paper emphasizes that the primary benefit was driven by mortality reduction and that the 6-month benefit was consistent with the trend seen in the raw mRS proportional-odds analysis. Note: 63 of 112 patients had strokes only in the MCA territory, 29 also involved ACA territory, and 7 also involved PCA territory; ~35% were in the dominant hemisphere.
Criticisms
- Small sample (N=112) and early stopping for efficacy can inflate estimated treatment effects; wide 95% CI on the primary OR (1.06-7.49) reflects imprecision.
- Open-label trial: outcome assessors were blinded but treating teams and patients/families were not, potentially influencing supportive-care decisions and withdrawal-of-care patterns.
- Primary endpoint of mRS 0-4 includes moderately severe disability (dependent for most bodily needs) as a 'good' outcome — many clinicians and patients would consider mRS 4-5 an unacceptable outcome; the dichotomy mRS 0-3 vs 4-6 was NOT significant.
- No survivor achieved mRS 0-2 in either arm; the benefit is essentially conversion of death to survival with severe disability.
- Depression, SF-36, and EQ-5D data available in only ~57% of survivors due to aphasia/severe cognitive impairment — introduces selection bias and limits interpretation of quality-of-life claims.
- Retrospective informed consent from a relative or physician was used in ~61-73% of enrollments — reflects the ethical challenge of studying obtunded patients but complicates generalization.
- 13 German sites only; homogeneous care setting may not generalize to systems with different ICU thresholds for withdrawal of care.
- No prespecified subgroup by hemisphere dominance, age band, infarct extent, or time-to-surgery reported in the main paper.
Funding
Deutsche Forschungsgemeinschaft (German Research Foundation), grant JU2764/1-1. Investigator-initiated; no commercial support.
Based on: DESTINY II (N Engl J Med, 2014)
Authors: Jüttler E, Unterberg A, Woitzik J, et al; DESTINY II Investigators
Citation: N Engl J Med 2014;370(12):1091-1100. DOI: 10.1056/NEJMoa1311367
Content summarized and formatted by NeuroTrials.ai.