Intensive Blood Pressure Lowering in Patients with Moderate to Severe Grade Acute Cerebral Hemorrhage: Post Hoc Analysis of Antihypertensive Treatment of Acute Cerebral Hemorrhage (ATACH)-2 Trial

The Antihypertensive Treatment of Acute Cerebral Hemorrhage (ATACH)-2 trial [1, 2] found that the treatment of participants with intracerebral hemorrhage (ICH) to achieve a target systolic blood pressure (SBP) of 110–139 mm Hg did not result in a lower rate of death or disability than standard reduction to a target of 140–179 mm Hg. The primary outcome of death or disability was observed in 38.7% of the participants (186 of 481) in the intensive-treatment group and in 37.7% (181 of 480) in the standard-treatment group (adjusted relative risk [RR]: 1.04; 95% confidence interval [CI]: 0.85–1.27). It is possible that an asymmetrically high proportion of patients with high Glasgow Coma Scale (GCS) scores can predispose the overall patient population to have a high rate of favorable outcome regardless of treatment (ceiling effect), making it difficult to discern the beneficial effect of intensive SBP reduction [2]. The observed rate (37.7%) of death or disability at 3 months was lower than the rates (60%) anticipated in trial design based on previous literature [2]. The lower than expected rate of death or disability was probably secondary to high proportion of patients with favorable baseline characteristics (e.g., 56% with a baseline GCS score of 15) [2]. We performed this analysis, which was not prespecified, to study the effect of intensive blood pressure reduction in patients with moderate to severe ICH within the subjects recruited in ATACH-2 trial.

The ATACH-2 trial was a randomized, multicenter, 2-group, open-label trial to determine the relative efficacy of intensive versus standard antihypertensive treatment that was initiated within 4.5 h after symptom onset and continued for the next 24 h in patients with spontaneous supratentorial ICH. The details are provided in previous publications [2, 3]. Patients aged ≥18 years with a GCS score of 5 or more and with a measurement of the intraparenchymal hematoma of <60 mL on initial computed tomographic (CT) scan were eligible for inclusion. Patients with intraventricular hemorrhage (IVH) associated with intraparenchymal hemorrhage and blood completely filling one lateral ventricle or more than half of both ventricles on initial CT scan were excluded. The goal of treatment was to reduce and maintain the hourly minimum SBP in the range of 140–179 mm Hg in the standard SBP reduction group and in the range of 110–139 mm Hg in the intensive SBP reduction group throughout the period of 24 h after randomization.

Baseline and 24-h CT scans (required by study) were forwarded to the core image analysis center. Serious adverse events were systematically reported up to 3 months after randomization. Nonserious adverse events were systematically reported up to day 7 or hospital discharge, whichever came first. Follow-up after discharge included telephone contact at 1 month and in-person clinical evaluation at 3 months. The data collection at the 3-month visit consisted of the score on the modified Rankin scale (mRs), quality of life as assessed by means of the European Quality of Life-5 Dimensions (EQ-5D) questionnaire [4], serious adverse events, and results of physical and neurologic examinations. The mRs grades the degree of disability or dependence in the daily activities using scores ranging from 0 (no symptoms) to 6 (death). The primary outcome was defined as a score of 4–6 [2]. The EQ-5D utility index (with scores ranging from −0.109 [least favorable health state] to 1 [most favorable health state], with 0 imputed for death) was derived by applying Shaw’s weight [5] to the response combinations of 5 questions regarding mobility, self-care, usual activities, pain and discomfort, and anxiety and depression.

The primary outcome was the proportion of patients who had moderately severe or severe disability or who had died (mRs score, 4–6) at 3 months. Hematoma expansion, defined as an increase of 33% or more in the hematoma volume from baseline to 24 h, was ascertained by central measurements. Safety outcomes were neurologic deterioration, defined as a decrease from baseline of 2 or more points in the GCS score or an increase of 4 or more points in the score on the National Institutes of Health Stroke Scale (NIHSS) that was not related to sedation or hypnotic-agent use and was sustained for at least 8 h within the 24 h after randomization; serious adverse events occurring within 72 h after randomization that were considered by the site investigator to be related to treatment; and death within 3 months after randomization.

Patients with moderate to severe grade ICH were identified based on previously published criteria [6, 7]: baseline GCS score <13 or NIHSS score ≥10; baseline intraparenchymal hemorrhage volume ≥30 mL; or presence of IVH. Patients who did not meet any of those criteria were classified as mild grade.

We compared demographic and clinical characteristics, frequency of hematoma expansion, neurological deterioration, serious adverse events, treatment-related adverse events, serious hypotension, and 3-month death or disability between patients classified as mild grade and those with moderate to severe grade ICH. We subsequently categorized the patients classified as mild grade and those with moderate severe grade ICH according to treatment group randomized. We used the χ² test, Fisher’s exact test, and ANOVA for categorical and the Wilcoxon 2-sample test for continuous variables. We subsequently compared the frequency of hematoma expansion, neurological deterioration, serious adverse events, treatment-related adverse events, serious hypotension, and 3-month death or disability between the 2 groups. Since the EQ-5D utility index scale scores were continuous variables, beta is the regression coefficient for the treatment effect (standard-treatment group divided by intensive-treatment group) in the generalized linear model. A beta of 0 indicates there was no effect of the treatment on the outcome. We tested the prognostic value of moderate to severe grade ICH using a logistic regression model including all patients with death or disability as the dependent variable. We tested the effect of death or disability by entering the interaction between treatment group (intensive vs. standard SBP reduction) and severity (moderate to severe vs. mild grade) in the model. We performed 2 logistic regression analyses to identify the effect of treatment group (intensive vs. standard SBP reduction) on (1) hematoma expansion and (2) death or disability after adjusting for age, gender, baseline NIHSS score, initial hematoma volume, and hematoma location separately for patients with moderate to severe grade and those with mild grade ICH. All analysis was performed using SAS 3.8 (Enterprise Edition) software.

Of a total of 1,000 subjects in Table 1, in 682 subjects who met the definition of moderate to severe grade (mean age 61.9 ± 13.1 years, 62.5% men) with a mean (standard deviation) baseline SBP of 174.7 (24.8) mm Hg, compared with 318 subjects with mild grade, there were no differences in mean age, race, ethnicity, and initial SBP. There was a higher proportion of subjects with basal ganglionic hemorrhages and a lower proportion of lobar hemorrhages in subjects with moderate to severe grade ICH. In Table 2, The primary endpoint of death or disability was observed in 50.7% (333/657) of subjects of moderate to severe grade, which was significantly higher than 11.2% (34/304) of subjects of mild grade (RR: 4.5; 95% CI: 3.3–6.3; p < 0.0001) with complete follow-up. The frequency of hematoma expansion within 24 h was not significantly higher (p = 0.09) among subjects of moderate to severe grade (24.2 and 19.4% for moderate to severe grade and mild grade groups, respectively). The rate of neurologic deterioration within 24 h was significantly higher (RR: 2.3; 95% CI: 1.4–3.9) among subjects of moderate to severe grade ICH (11.6%) than among subjects of mild grade (5.0%). The percentage of patients with treatment-related serious adverse events within 72 h was 1.5% in the moderate-to-severe grade group and 1.3% in the mild grade group. Serious adverse events were significantly higher (RR: 2.5; 95% CI: 1.8–3.4) among patients in the moderate-to-severe grade group (28.0 and 11.3% for moderate to severe grade and mild grade, respectively). The median values of the EQ-5D utility index score and EQ-5D visual-analog scale score were significantly lower among subjects with moderate to severe grade ICH.

Among subjects with moderate to severe grade ICH in Table 3, 336 were assigned to intensive SBP reduction and 346 to standard SBP reduction. The primary endpoint of death or disability was observed in 52.5% (170/324) of subjects receiving intensive SBP reduction and 48.9% (163/333) of subjects receiving standard SBP reduction (RR: 1.1; 95% CI: 0.9–1.2) with complete follow-up. The frequency of hematoma expansion within 24 h was significantly lower (RR: 0.7; 95% CI: 0.55–0.96) among subjects randomized to intensive SBP reduction (20.4 and 27.9% for intensive and standard SBP reduction groups, respectively). There was no difference in the rates of neurologic deterioration at 24 h after randomization between the 2 groups. The percentage of patients with treatment-related serious adverse events within 72 h after randomization was 1.8% in the intensive-treatment group and 1.2% in the standard-treatment group. Serious adverse events were significantly higher (RR: 1.4; 95% CI: 1.1–1.8) among patients randomized to intensive SBP reduction (32.7 and 23.4% for intensive and standard SBP reduction groups, respectively). There were no significant between-group differences in the rate of death at 3 months. In logistic regression analysis, intensive SBP reduction was significantly associated with hematoma expansion (odds ratio [OR]: 0.6; 95% CI: 0.44–0.92) after adjusting for other potential confounders only for patients with moderate to severe grade ICH. There was no association between intensive SBP reduction and death or disability in moderate to severe grade ICH.

Among subjects with mild grade ICH in Table 4, 164 were assigned to intensive treatment and 154 to standard treatment. The primary endpoint of death or disability was observed in 10.2% (16/157) of subjects receiving intensive SBP reduction and 12.2% (18/147) of subjects receiving standard SBP reduction (RR: 0.8; 95% CI: 0.4–1.6) with complete follow-up. The frequency of hematoma expansion within 24 h was not different (RR: 0.8; 95% CI: 0.5–1.3) among subjects randomized to intensive SBP reduction (17.3 and 21.7% for intensive and standard SBP reduction groups, respectively). There was no difference in the rates of neurologic deterioration at 24 h and treatment-related serious adverse events within 72 h after randomization between the 2 groups. There were also no significant between-group differences in the rates of serious adverse events or death at 3 months. There was no association between intensive SBP reduction and hematoma expansion or death or disability in the logistic regression analyses after adjusting for other potential confounders in patients with mild grade ICH.

In a multivariate model including all randomized patients (n = 961), moderate to severe grade ICH was a predictor of death or disability at 3 months (OR: 9.57; 95% CI: 6.37–14.33). Age also was a predictor of death or disability at 3 months (OR: 1.05; 95% CI: 1.04–1.06). The interaction between treatment group (intensive vs. standard SBP reduction) and severity (moderate to severe vs. mild grade) was not significant (p = 0.33).

We found that patients with moderate to severe grade ICH had higher rate of death or disability at 3 months and higher frequency of hematoma expansion compared with patients with mild grade ICH. Intensive SBP reduction reduced the frequency of hematoma expansion in patients with moderate to severe grade ICH but failed to reduce the death or disability compared with standard SBP reduction. The lack of benefit with intensive SBP reduction was further confirmed by absence of any modifying effect (nonsignificant interaction) between the SBP reduction-treatment group and the ICH severity group in determining death or disability at 3 months after randomization.

The beneficial effect of intensive SBP reduction was expected to be prominent in patients with moderate to severe grade ICH. Initial volume of hematoma has a direct relationship with subsequent hematoma expansion with larger hematomas at baseline correlating with higher frequency of expansion [8-11]. The combined analysis of INTERACT 1 and 2 studies [11] suggested that ICH volume of >20 mL (OR: 5.57) and 10–20 mL (OR: 2.89) was associated with increased risk of hematoma enlargement (compared with those with hematoma volume of <10 mL). Clinical severity measures such as GCS or NIHSS scores are shown to be directly associated with surrogate markers of continued bleeding such as contrast extravasation seen on CT angiography and other hematoma expansion parameters in multiple studies [12-14]. In a systematic review of 36 cohorts consisting of 5,435 eligible patients [13], baseline GCS scores of 7–12 (OR: 2.83) and 13–14 (OR: 1.99) had higher frequency of hematoma expansion compared with those with a GCS score of 15. ICH volume on baseline imaging was the strongest predictor of hematoma enlargement (OR: 7.18). The relationship between IVH and hematoma enlargement is less clear. The combined analysis of INTERACT 1 and 2 studies [11] suggested that IVH was associated with increased risk of hematoma enlargement (OR: 2.04). Some studies have identified no relationship [14, 15] or even that IVH was associated with decreased risk of hematoma enlargement [16]. The ATACH-2 study design had excluded patients with very high anticipated early mortality to avoid a “floor effect” meaning that due to severity of initial injury will overshadow any therapeutic intervention [1-3]. Patients with a GCS score of <5, those with a measurement of the intraparenchymal hematoma of ≥60 mL, or those in whom IVH was associated with intraparenchymal hemorrhage and blood completely filled one lateral ventricle or more than half of both ventricles on initial CT scan were excluded. The Factor Seven for Acute Hemorrhagic Stroke (FAST) subgroup analysis [17] had suggested that the effect of reduction in hematoma expansion (using rFVIIa treatment) on clinical outcome was most prominent in patients with baseline hematoma <60 mL and baseline IVH volume <5 mL. Other factors such as age <70 years and time interval between symptom onset and randomization ≤2.5 h also increased the benefit of hematoma reduction in the FAST subgroup analysis. Therefore, since the postulated mechanism of intensive SBP reduction is by ameliorating hematoma expansion, a greater magnitude of effect is expected in patients with larger baseline hematomas, those with lower GCS or higher NIHSS score, and perhaps those with IVH.

As expected, based on the literature cited above, the frequency of hematoma expansion was significantly lower among moderate to severe grade ICH patients randomized to intensive SBP reduction (20.4 and 27.9% for intensive and standard SBP reduction groups, respectively [an absolute decrease of 7.5%]). In patients with mild grade ICH, the frequency of hematoma expansion was not significantly different among patients randomized to intensive SBP reduction (an increase of 4.4%). However, the rate of death or disability was not different between patients randomized to intensive SBP reduction compared with those randomized to standard SBP reduction in either moderate to severe grade or mild grade ICH. In the primary analysis of ATACH-2, which included all randomized patients, the frequency of hematoma expansion was significantly different between those randomized to intensive SBP reduction compared with those randomized to standard SBP reduction (19.3 vs. 26.0%, absolute decrease of 6.7%; RR: 0.7; 95% CI: 0.6–0.9). The rate of death or disability was not different between patients randomized to intensive SBP reduction compared with those randomized to standard SBP reduction in primary analysis of ATACH-2. Our results highlight the discrepancy between reduction in hematoma expansion and lack of clinical benefit in terms of death or disability. A pooled individual patient meta-analysis [7] using data from 218 placebo-treated patients in recombinant activated factor VII (rFVIIa) trials and 103 patients in the Cincinnati observation study quantified the benefit of reduction in hematoma expansion on patient outcomes in ICH. The analysis concluded that hematoma growth was a determinant of mortality and poor outcome after ICH [7]. Percentage hematoma growth predicted both mortality and mRs of 4–6 at 3 months following ICH occurrence. An analysis of 531 patients with ICH from the Virtual International Stroke Trials Archive [18] found that all definitions of hematoma expansion predicted various definitions of poor outcome. The highest positive predictive values were seen when using an absolute growth definition based on absolute volume change to predict more severe outcomes (mRs of 5–6). In the pilot INTERACT study [19], a 10.7-mL (1-SD) increase in hematoma volume over 24 h was strongly associated with poor outcome (adjusted odds ratio [OR]: 1.72). A 1-mL increase in hematoma growth was associated with a 5% (95% CI: 2–9%) higher risk of death or dependency.

One of the first evidence of discrepancy between reduction in hematoma expansion and lack of effect on reduction in death or disability was seen in the FAST trial [20]. Treatment with 80 μg/kg of rFVIIa resulted in a significant reduction in growth in volume of the hemorrhage (mean estimated increase in volume of the ICH of 11% compared with 26% in the placebo group). However, there was no significant difference among the 2 groups in the proportion of patients with poor clinical outcome (mRs of 5–6 at 3 months; 24% in the placebo group and 29% in the group receiving 80 μg/kg). In the pilot INTERACT study [21], ICH patients who were randomized to intensive reduction of BP (target systolic BP 140 mm Hg; n = 203) had lower frequency of hematoma expansion compared with standard guideline-based management of BP (target systolic BP 180 mm Hg; n = 201). The frequency of hematoma expansion (an increase in hematoma volume of >33% or >12.5 mL during the first 24 h after ICH onset) was 15% in the intensive SBP reduction group compared with 22% in the standard SBP reduction group (p < 0.07). However, the primary outcome of death or dependency (mRs of 3–6) at months was not different between the 2 groups (49 vs. 48%). Furthermore, in Tranexamic acid for hyperacute primary IntraCerebral Haemorrhage (TICH-2) trial [22] 2,325 subjects were randomized to receive either 1-g intravenous tranexamic acid bolus followed by an 8-h infusion of 1-g tranexamic acid or a matching placebo, within 8 h of symptom onset. The frequency of hematoma expansion at day 2 was lower in the tranexamic acid group (25 vs. 29%, p = 0.03). However, functional status defined by mRs at day 90 did not differ significantly between the groups in ordinal regression analysis after adjustment for the stratification and minimization criteria (aOR: 0.88; 95% CI: 0.76–1.03; p = 0.11).

One question is whether mRs is sensitive enough to identify clinical benefit of reduction in frequency of hematoma expansion with an intervention. mRs evaluates the domains of mobility, self-care, and independence but not cognitive or social functions [23]. New patient-reported outcome measures (PROMs) such as Patient-Reported Outcome Measurement Information System (PROMIS), Quality of Life in Neurological Disorders (NeuroQOL), and 10-item PROMIS Global Health (PROMIS-10) [23-25] may provide more sensitive assessment of clinical benefit on reduction in frequency of hematoma expansion in future clinical trials. There are new data that demonstrate improvement in mRs grades in ICH patients across the disability spectrum from 3 to 6 months and perhaps even after 6 months in patients who are less disabled [26]. Whether ascertaining the primary endpoint at 6 or 12 months (rather than 3 months) after randomization is more likely to demonstrate the clinical benefit of reduction in frequency of hematoma expansion remains unknown. Recent trials such as Minimally Invasive Surgery Plus Alteplase for Intracerebral Hemorrhage Evacuation (MISTIE) III have used mRs ascertained at 12 months as primary outcome to capture the late improvements in disability [27]. The analysis had the limitations of post hoc analysis [28-30]. Most authors agree that post hoc or subgroup analysis can examine subsets of the data accrued in the setting of a clinical trial in order to answer clinically relevant questions that were not addressed in the prespecified data analysis plan since it is not practical to design a randomized clinical trial for every clinical question or population of interest. Such analysis may identify groups of patients who may respond differently to treatment than other groups. However, results of post hoc analysis may be contaminated by chance, other patient differences, and different co-interventions. We acknowledge that the subgroup hypothesis or the direction of effect was not prespecified [28, 29] for this analysis. The small number of patients in subgroups may not allow identification of smaller differences in outcome between the subgroups, a type II error [31]. Therefore, these are exploratory data analyses for hypothesis generation, and extrapolation of these results to individual patients may be premature.

In conclusion, intensive SBP reduction reduced the frequency of hematoma expansion but did not reduce the rate of death of death or disability in patients with moderate to severe grade ICH.

The protocol and consent forms were approved by the institutional review board or equivalent ethics committee at each participating site, and all participants or their legally authorized representative provided written informed consent before randomization.

The authors have no conflicts of interest to declare.

This study was funded by the National Institute of Neurological Disorders and Stroke and the National Cerebral and Cardiovascular Center; ATACH-2 ClinicalTrials.gov number: NCT01176565.

All authors substantially contributed to the conception and design of the work, the acquisition, analysis, and interpretation of data for the work, and drafting the work and revising it critically for important intellectual content. All authors gave final approval of the version to be published and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.