ATACH-2 - Intensive BP lowering in ICH

Qureshi AI, et al. Intensive blood-pressure lowering in patients with acute cerebral hemorrhage. NEJM 2016;online


Clinical Question

In patients with intracerebral hemorrhage (ICH), does an SBP goal 110-140 mmHg improve functional outcomes compared to an SBP goal of 140-179 mmHg?


Bottom Line

In patients with ICH, more intensive BP control did not improve functional outcomes. Based on the results of this study, SBP should be maintained ~140-180 mmHg.



Allocation-concealed, open-label RCT, 4% loss-to-follow-up, analyzed using the intention-to-treat population.


  • The trial was truncated (stopped early) after the 2nd pre-specified interim analysis for futility.
  • A central radiologist blind to treatment allocation read the baseline & 24h head CT to determine hematoma expansion.
  • The 3-month outcome assessor was not involved in the randomization or inpatient management of participants, but was not specifically blinded to treatment allocation.


Patients and Setting

  • 6 countries (China, Germany, Japan, South Korea, Taiwan, USA), 110 centers
  • May 2011 - September 2015
  • Inclusion criteria:
    1. Men or women 18+ y
    2. Intracerebral hemorrhage, as defined by
      • Clinical signs consistent with dx of stroke
      • Head CT demonstrates intraparenchymal hematoma, volume <60 mL
    3. Initial NIHSS score 4+ (/42)
    4. GCS 5+ (/15)
    5. Onset of new neuro deficits within 2.5 h
    6. Admission SBP 180-240 mmHg on 2 repeat measurements >5 min apart, but <10 min apart
    7. IV nicardipine could be started within 4.5 h of symptom onset
  • Key exclusion criteria:
    • Considered a candidate for immediate neurosurgical intervention (e.g. ventriculostomy insertion, evacuation)
    • Secondary causes of ICH
      • Previously known neoplasms, AVM, aneurysms
      • traumatic ICH
    • Infratentorial ICH (e.g. pons, cerebellum)
    • Presence of intraventricular hemorrhage (IVH) that completely fills 1 lateral ventricule or >1/2 of both ventricles
    • Pre-morbid modified Rankin Scale (mRS) score 4+ (/6)
    • Increased risk of bleeding
      • Hx of bleeding diathesis or coagulopathy
      • Warfarin use in last 5 days
      • Platelets <50
  • 8,532 screened -> 1,000 randomized -> 961 analyzed @ 3 months
  • Average patient:
    • 62 y
    • 38% female
    • Race: 56% Asian, 28% White, 12% Black
    • PMHx
      • Prior stroke/TIA 16%
      • HTN 80%, previous use of BP meds 50%)
    • GCS: 15 (55%), 12-14 (30%), 3-11 (15%)
    • Mean SBP at presentation 200 mmHg
    • Median NIHSS score 11
    • ICH volume
      • >30 cm^3 (10%)
      • Median 10 cm^3
    • IVH 25%
    • Hemorrhage location: Basal ganglia (51%), thalamus (37%), lobar (10%)


Intervention and Control

  • Intervention: Target SBP 110-139 mmHg x24h 
    • 71% started <3h of symptom onset, 29% started 3-4.5h after symptom onset
    • Mean minimum SBP during initial 2h of starting tx: 129 mmHg
    • Reached target SBP within 2h of starting tx: 88%
  • Control: Target SBP 140-179 mmHg x24h
    • 64% started <3h of symptom onset, 36% started 3-4.5h after symptom onset
    • Mean minimum SBP during initial 2h of starting tx: 141 mmHg
    • Reached target SBP within 2h of starting tx: 99.2%
  • Co-interventions common to both groups:
    • Adjunctive care according to the American Stroke Association guidelines
    • Mean time between symptom onset & randomization: ~3 hours in both groups
    • 1st-line antihypertensive: Nicardipine IV
      • Initial dose of 5 mg/h
      • Titrated q15min by 2.5 mg/h
      • Max rate 15 mg/h
    • 2nd-line: Labetalol IV (used if SBP higher than target despite max nicardipine rate x30 min)
      • Diltiazem or urapidil IV used in countries where labetalol was not available
    • CT head, non-contrast @ 24h after start of treatment



ATACH-2 Outcomes

ATACH-2 organ-specific SAEs

  • Long-term outcomes (@ 3 months):
    • No statistically significant difference in death or disability (modified Rankin Scale [mRS] 4-6), or in any mRS score.
    • No statistically significant difference in median quality of life scores, with wide variation from terrible to favorable health state in each group.
    • Possible 5.6% absolute risk increase in serious adverse events (SAEs) in the intensive BP control group (NNH 18). This increased risk in overall SAEs isn't explained by increase in any organ-specific SAE, and is therefore difficult to accept as a real effect.
  • Short-term outcomes
    • Neurological deterioration within 24h of treatment onset (reminder: the randomized treatment target was maintained for 24h) was not statistically significant between groups
    • The proportion of patients with hematoma expansion >33% was not statistically significantly different between groups, though the confidence interval could not rule out a difference of up to 10% in favor of the intensive treatment group. Still, this is only a surrogate for long-term functional outcome, which was not improved in this trial.
    • Hypotension during the treatment phase was rare and similar between groups (~1%).
  • Subgroup analyses: Intensive BP control was not statistically significantly better (or worse) in any of the subgroup analyses performed (sex, race, geography, T2DM, GCS, hematoma volume or location, IVH, meeting/not meeting SBP target at 2h).

Key Considerations


  • P: ATACH-2 participants appear to be representative of patients with ICH seen in real-world practice. Notably, however, only 1/8 of patients considered for enrollment were randomized, with no further definition of reason for exclusion.
  • I/C: Notably, the BP-lowering intervention was delivered quickly & effectively, with 88% of patients in the intervention group achieving target SBP, & near all in the control group achieving target SBP. There was also an early, large separation in SBP between groups. The lack of benefit from intensive BP control cannot therefore be explained by inability to achieve this BP goal.
  • O: The long-term outcomes of this trial, including mRS, quality of life measures and serious adverse events, are patient-centered and highly important.

Internal validity:

  • The open-label nature of the study was unavoidable. This "opens" the study up to performance and detection bias, though the risk for these is likely low in this context.
    • Performance bias is unlikely given the otherwise standardized care processes in ICH management.
    • Detection bias could be problematic, especially given that the outcome assessor was unlikely to be blinded. None of the objective outcomes (death) or "relatively" objective outcomes (moderate-severe, severe disability on the mRS; overall serious adverse events) favored intensive BP control.


How To Reconcile the Results of ATACH-2 and INTERACT2

Summary of INTERACT2:

  • P: Enrolled 2,839 patients with spontaneous ICH within 6h of symptom onset with GCS 6-15 & baseline SBP 150-220 mmHg in whom neurosurgical intervention was not planned
    • Average patient: 63.5 y, 2/3 male, median GCS 14, median NIHSS score 10-11, use of antiplatelet/oral anticoagulant ~10%, median baseline ICH volume 11 mL, ~84% located in basal ganglia or thalamus, 28% with IVH
  • I/C: Assigned to target SBP <140 mmHg or <180 mmHg to be achieved within 1h and maintained x7 days, starting with locally-available IV antihypertensives
    • IV antihypertensive used in 90% of target SBP <140 mmHg group vs 43% of SBP <180 mm Hg group
    • Wide variety of agent used, including alpha1-antagonists (32% vs 13%), CCB such as nicardipine or nimodipine (16% vs 8%), labetalol (14% vs 6%), or others
    • Mean SBP achieved at 1h were 150 mmHg in the intensive group (33% achieved target) & 164 mmHg in the control group.
  • O:
    • Non-significant reduction in primary outcome of mRS 3-6 with intensive BP lowering (52.0% vs 55.6%, p=0.06) that was statistically-significant using a secondary analysis method (ordinal analysis p=0.04)
    • Statistically significant reduction in EQ-5D utility index (0.60 vs 0.55, p=0.002) and certain individual quality of life measures
    • No statistically significant difference in any other outcome, including any specific mRS category, serious adverse events, duration of hospital stay, neuro deterioration <24h, hematoma growth, and hypotension.

Key similarities & differences between ATACH-2 and INTERACT2:

  • P
    • Enrolled patients with similar severity of illness, hematoma size & location
    • Baseline SBP higher in ATACH-2 vs INTERACT2 (~200 vs 180 mmHg) due to higher requirement for inclusion (180-240 mmHg vs 150-220 mmHg)
  • I
    • ATACH-2 participants achieved target SBP more quickly & more frequently than in INTERACT2
      • In fact, achieved BP in INTERACT2 intervention arm closer to ATACH-2 control arm (mean SBP @ 1h 150 mm Hg & minimum SBP <1h 141 mm Hg, respectively)
    • ATACH-2 used nicardipine IV infusion as 1st line in all patients, whereas used infrequently in INTERACT2, with more frequent use of intermittent boluses of alpha1-antagonists and other agents
    • The intervention duration in ATACH-2 was limited to the first 24h, whereas INTERACT2 targeted both acute (first 24h) and subacute (post-admission day 1-7) BP control.
  • O
    • Similar outcome measures in both trials with slightly different definitions of "bad" functional outcome (mRS 4-6 in ATACH-2 versus 3-6 in INTERACT2)
    • Both trials showed no statistically significant difference in functional outcome in the primary analysis, & no clinically significant difference in any single mRS category
    • INTERACT2 showed statistically significant improvement in long-term quality of life with intensive BP control, whereas ATACH-2 did not.


Summary author: Ricky Turgeon BSc(Pharm), ACPR, PharmD
Summary date: 23 June 2016