Abstract
Background: It is unknown if race/ethnicity modifies the response to blood pressure (BP) lowering treatment after intracerebral hemorrhage (ICH). We aimed to examine the race/ethnicity differences in the response to BP lowering treatment after ICH. Methods: This is a post hoc analysis of the Antihypertensive Treatment of Acute Cerebral Hemorrhage II (ATACH-2) trial. The primary outcome is good outcome, defined as 90-day modified Rankin Scale 0–3. The primary predictor is race/ethnicity for which we included non-Hispanic categories of White, Black, Asian, and the category of Hispanic. We fit adjusted logistic regression models with the predictor of race/ethnicity and models with the interaction term of treatment*race/ethnicity. Results: We included a total of 953 patients in our analysis (White = 213, Black = 112, Asian = 554, and Hispanic = 74). In the models with the interaction between race/ethnicity and treatment, we found that White patients assigned to the intensive treatment arm had lower predicted probability of good outcome than those assigned to the standard treatment arm (Model 1: 56.2% vs. 68.1%, p = .027; Model 2: 53.4% vs. 68.3%, p = .009). When divided into White and non-White groups, intensive treatment was associated with higher odds of serious adverse events in White group but not in the non-White group. In addition, there was an association between intensive treatment and higher risk of hematoma expansion in White patients and lower risk of hematoma expansion in non-White patients. Conclusions: In the ATACH-2, there was an interaction between race/ethnicity and response to BP lowering treatment after ICH, with White patients having an association between intensive blood pressure reduction and worse outcome.
Keywords: Stroke, blood pressure, hemorrhage, race
Introduction
The benefit of blood pressure (BP) lowering treatment after intracerebral hemorrhage (ICH) remains uncertain.1–4 The Antihypertensive Treatment of Acute Cerebral Hemorrhage II (ATACH-2) trial demonstrated no benefit of intensive BP lowering treatment after ICH. 2 A subsequent meta-analysis of randomized trials suggested a potential benefit of BP reduction after intracerebral hemorrhage. 3 Considering the available evidence, the European Stroke Organization (ESO) guidelines recommend lowering systolic BP to less than 140 mmHg in patients with intracerebral hemorrhage presenting within 6 h of symptoms onset (weak recommendation). 4
Previous studies suggested racial/ethnic variation in ambulatory BP trajectories, and response to antihypertensive medications.5–9 However, it is still unknown whether the response to BP lowering treatment after acute ICH differs according to race/ethnicity. In this post hoc analysis of the ATACH trial, we examined race/ethnicity differences and their relationship to BP lowering treatment after ICH.
Methods
This is an exploratory non-prespecified analysis of ATACH-2, a randomized trial of patients with non-traumatic ICH to determine if an intensive antihypertensive goal (goal SBP of 110–139 mm Hg) was superior to a standard goal (goal SBP of 140–179 mm Hg). 2 The anonymized ATACH-2 dataset, received with a local IRB waiver, is available through NINDS at https://www.ninds.nih.gov. The primary outcome is good outcome, defined as 90-day modified Rankin Scale (mRS) score of 0–3. Additional outcomes included neurologic deterioration, hematoma expansion, serious adverse events within 3 months, and withdrawal of care. Neurological deterioration was defined as a decrease from baseline of two or more points in Glasgow Coma Scale score or an increase of four or more points in the National Institutes of Health Stroke Scale score that was not associated with sedation or hypnotic agent use and was sustained for at least 8 h within the 24 h after randomization. Hematoma expansion was defined as an increase of 33% or more in the hematoma volume from baseline to 24 h 2
The primary predictor is race/ethnicity for which we included non-Hispanic categories of White, Black, Asian, and the category of Hispanic. We fit logistic regression models to our outcome and adjusted for 1) Model 1: covariates selected with a restricted LASSO methodology 10 including patient age, intubation, baseline NIH Stroke Scale, and baseline intraparenchymal and intraventricular hemorrhage volume; and 2) Model 2: a priori for patient age, sex, baseline Glasgow Coma Scale, baseline ICH score, hemorrhage location, hemisphere, intubation, surgical intervention, and baseline intraparenchymal and intraventricular hemorrhage volume. We verified the models were not multicollinear by confirming the variance inflation factor was <10. To show the impact of the ATACH-2 intervention of intensive in race categories, we included the interaction of treatment-race/ethnicity in the models and used marginal effects to derive the predicted probability of good outcome by race categories. We also stratified patient by White versus non-White. In a sensitivity analysis, we excluded patients who had withdrawal of care. The data that support the findings of this study are available from the corresponding author upon reasonable request.
Results
We included 953 patients in our analysis, with 39 patients excluded because of missing outcome data and eight patients excluded for having other racial categories. The baseline characteristics of the study groups are reported in Table 1. White patients were older, more likely to be female, had higher NIH stroke scale, and ICH score than the other groups. Baseline ICH volume was similar across groups, but intraventricular hemorrhage volume was higher in White patients. Baseline systolic BP and mean systolic BP in the first 24 h were highest in Black patients and lowest in White patients. Coronary artery disease and atrial fibrillation were more common in White patients than other groups. In contrast, baseline hypertension was less frequent in White patients. Smoking was significantly less frequent in White patients.
Table 1.
Baseline demographics by race/ethnicity category. a
| Variable | White (n=213) | Black (n=112) | Asian (n=554) | Hispanic (n=74) | p value |
|---|---|---|---|---|---|
| Age | 68.2 ± 12.7 | 57.6 ± 10.1 | 61.1 ± 13.1 | 58.3 ± 13.0 | <.001 |
| Female sex | 107 (50.2%) | 49 (43.8%) | 189 (34.1%) | 19 (25.7%) | <.001 |
| Hemorrhage location | <.001 | ||||
| Basal ganglia | 110 (51.6%) | 58 (52.3%) | 321 (57.9%) | 47 (64.4%) | |
| Thalamus | 62 (29.1%) | 32 (28.8%) | 195 (35.2%) | 18 (24.6%) | |
| Lobar | 41 (19.3%) | 21 (18.9%) | 38 (6.9%) | 8 (11.0%) | |
| Left hemisphere | 104 (48.8%) | 51 (45.5%) | 262 (47.3%) | 33 (44.6%) | .559 |
| Minutes from onset to randomization | 193.0 ± 55.6 | 197.5 ± 50.5 | 174.8 ± 57.4 | 203.0 ± 53.1 | <.001 |
| Baseline NIH Stroke Scale | 12, 6.5–17 | 11, 6–17.5 | 10, 6–15 | 11, 6–17.5 | .249 |
| Baseline Glasgow Coma Scale | 15, 13–15 | 15, 14–15 | 15, 13–15 | 15, 13–15 | .796 |
| ICH score | 1, 0–2 | 0, 0–1 | 0, 0–1 | 0, 0–1 | .004 |
| Baseline intracerebral hemorrhage volume (ml, n=939) | 17.0 ± 18.3 | 13.7 ± 13.2 | 14.5 ± 12.7 | 16.1 ± 16.2 | 0.122 |
| Baseline intraventricular hemorrhage volume (ml, n=939) | 4.3 ± 9.0 | 2.1 ± 4.9 | 1.8 ± 5.3 | 2.9 ± 6.7 | <.001 |
| Baseline systolic blood pressure (n=952) | 198.4 ± 29.0 | 207.9 ± 30.7 | 199.7 ± 24.5 | 202.2 ± 32.1 | 0.016 |
| Mean systolic blood pressure during trial intervention (n=952) | 141.1 ± 12.3 | 145.7 ± 11.7 | 141.9 ± 13.3 | 143.0 ± 12.1 | 0.013 |
| Intubation (n=947) | 46 (22.0%) | 23 (20.5%) | 39 (7.1%) | 17 (23.0%) | <.001 |
| External ventricular drain (n=947) | 24 (12.0%) | 9 (8.0%) | 23 (4.2%) | 11 (14.9%) | <.001 |
| Surgical decompression or evacuation (n=947) | 5 (2.4%) | 9 (8.0%) | 26 (4.7%) | 2 (2.7%) | .107 |
| Intensive treatment arm | 109 (51.2%) | 60 (53.6%) | 275 (49.6%) | 34 (46.0%) | .754 |
| Prior stroke (n=947) | 39 (18.6%) | 27 (24.1%) | 68 (12.3%) | 18 (25.0%) | .001 |
| Atrial fibrillation (n=943) | 22 (10.6%) | 2 (1.8%) | 9 (1.6%) | 2 (2.8%) | <.001 |
| Coronary artery disease (n=945) | 21 (10.1%) | 10 (9.0%) | 10 (1.8%) | 2 (2.8%) | <.001 |
| Diabetes (n=936) | 36 (17.4%) | 32 (28.8%) | 78 (14.3%) | 28 (39.4%) | <.001 |
| Hypertension (n=936) | 158 (76.3%) | 104 (93.7%) | 433 (80.5%) | 61 (84.7%) | .001 |
| Smoker | 27 (12.7%) | 30 (26.8%) | 174 (31.4%) | 16 (21.6%) | <.001 |
| Baseline glucose (n=945) | 140.4 ± 61.4 | 136.0 ± 55.5 | 135.3 ± 48.8 | 152.8 ± 70.9 | .062 |
| Baseline creatinine | 1.0 ± 0.5 | 1.4 ± 1.3 | 1.0 ± 1.0 | 1.3 ± 1.1 | <.001 |
| Baseline white blood cell count | 8.8 ± 3.1 | 8.0 ± 9.2 | 8.0 ± 2.8 | 8.5 ± 2.8 | .072 |
| Baseline platelet count | 224.0 ± 61.4 | 237.1 ± 78.5 | 215.2 ± 61.1 | 220.3 ± 48.2 | .006 |
| Intracerebral hemorrhage growth at 24 h (ml, n=939) | 0.03 ± 15.7 | 1.5 ± 8.3 | 0.7 ± 11.7 | −1.3 ± 12.7 | .453 |
| Hematoma expansion ≥33% (n=934) | 30 (14.5%) | 17 (15.6%) | 87 (15.9%) | 9 (12.7%) | .887 |
| Hematoma expansion >6 ml (n=934) | 26 (12.6%) | 15 (13.8%) | 64 (11.7%) | 7 (9.9%) | .864 |
| Death | 40 (18.8%) | 8 (7.1%) | 11 (2.0%) | 7 (9.5%) | <.001 |
| Withdrawal of care | 29 (13.6%) | 5 (4.5%) | 6 (1.1%) | 5 (6.8%) | <.001 |
| Neurologic deterioration within 24 h | 29 (13.6%) | 15 (13.4%) | 38 (6.9%) | 8 (10.8%) | .012 |
| Serious adverse event within 3 months | 82 (38.5%) | 39 (34.8%) | 78 (14.1%) | 17 (23.0%) | <.001 |
| Good outcome | 109 (51.2%) | 61 (54.5%) | 373 (67.3%) | 45 (60.8%) | <.001 |
| Good outcome | |||||
| Intensive arm | 53 (48.6%) | 30 (50.0%) | 189 (68.7%) | 21 (61.8%) | .001 |
| Standard arm | 56 (53.9%) | 31 (59.6%) | 184 (66.0%) | 24 (60.0%) | .173 |
aWhite, Asian, and Black are non-Hispanic. Binary variables presented as n (%); ordinal as median, IQR; and interval as mean ± SD; statistical significance tested with the chi-squared test for binary variables, Kruskal–Wallis for ordinal variables, and ANOVA for interval variables.
Good outcome was achieved in 51.2%, 54.5%, 67.3%, and 60.8% in Whites, Blacks, Asians, and Hispanics, respectively (Table 1). The mean variance inflation factor for Model 1 was 3.17 and for Model 2 was 8.67, indicating acceptable multicollinearity. In Model 1, the predicted probability of good outcome by race category was 62.0%, 52.5%, 63.4%, and 63.7% in Whites, Blacks, Asians, and Hispanics, respectively.
Because the adjusted probability of good outcome was lower for Black patients, we compared Black patients to all other race categories and found their odds ratio for good outcome in Model 1 was 0.50 (95% CI 0.29–0.84, p = .009) and in Model 2 was 0.46 (95% CI 0.27–0.75, p = .002).
After introducing the treatment*race interaction in the adjusted model, we found that White patients had a significantly lower predicted probability of good outcome with intensive treatment compared to standard treatment (Model 1: 56.2% vs. 68.1%, p = .027; Model 2: 54.0% vs. 67.8%, p = .015). There were not significant differences for other race categories (Table 2). In logistic regression adjusted identically to the main models but only including White patients, the odds ratio for good outcome with intensive treatment in Model 1 was 0.46 (95% CI 0.21–0.98, p = .045) and in Model 2 was 0.37 (95% CI 0.17–0.78, p = .010). In a sensitivity analysis excluding 47 patients with withdrawal of care, our findings remained consistent, with even larger effect sizes for some comparisons (data not shown).
Table 2.
Predicted probability of good outcome with the interaction between race and treatment arm.
| Model 1 | Model 2 | |||||
|---|---|---|---|---|---|---|
| Race a | Predicted probability of good outcome b | 95% CI | p value | Predicted probability of good outcome c | 95% CI | p value |
| White | .027 | .015 | ||||
| Standard BP | 68.1% | 60.9–75.3 | 67.8% | 60.2–75.5 | ||
| Intensive BP | 56.2% | 48.3–64.1 | 54.0% | 45.7–61.7 | ||
| Black | .954 | .776 | ||||
| Standard BP | 52.3% | 41.2–63.4 | 51.3% | 39.7–62.9 | ||
| Intensive BP | 52.7% | 42.3–63.2 | 48.9% | 37.8–60.1 | ||
| Asian | .493 | .567 | ||||
| Standard BP | 62.2% | 57.8–66.7 | 63.2% | 58.7–67.8 | ||
| Intensive BP | 64.4% | 60.1–68.7 | 65.1% | 60.6–69.5 | ||
| Hispanic | .318 | .433 | ||||
| Standard BP | 59.6% | 47.1–72.2 | 57.2% | 43.9–70.6 | ||
| Intensive BP | 68.6% | 56.1–81.2 | 65.2% | 50.4–80.0 | ||
aWhite, Asian, and Black are non-Hispanic.
bn=931, adjusted for patient age, intubation, baseline NIH Stroke Scale, and baseline intraparenchymal and intraventricular hemorrhage volume.
cn=931, adjusted for patient age, sex, baseline Glasgow Coma Scale, hemorrhage location, hemisphere, intubation, surgical intervention, and baseline intraparenchymal and intraventricular hemorrhage volume.
Subsequently, we stratified patients into White and non-White categories to look at the secondary outcomes. In White patients, intensive treatment was associated with lower odds of good outcome and higher odds of hematoma expansion and serious adverse events (Table 3). Conversely in non-White patients, intensive treatment was associated with lower odds of hematoma expansion (Table 3).
Table 3.
Outcomes in white and non-white patients, according to treatment groups.
| Outcome | Intensive treatment (n=109) | Standard treatment (n=104) | Odds ratio a (95% CI) for intensive treatment | p value |
|---|---|---|---|---|
| White patients | ||||
| Good outcome | 53/109, 48.6% | 56/104, 53.9% | 0.46 (0.21–0.98) | .045 |
| Death | 19/109, 17.4% | 21/104, 20.2% | 2.08 (0.73–5.96) | .171 |
| Hematoma expansion | 20/104, 19.2% | 10/103, 9.7% | 2.53 (1.07–5.98) | .035 |
| Neurologic deterioration within 24 h | 17/109, 15.6% | 12/104, 11.5% | 2.71 (0.99–7.38) | .052 |
| Serious adverse event within 3 months | 46/109, 42.2% | 36, 34.6% | 2.46 (1.20–5.05) | .014 |
| Withdrawal of care | 13/109, 11.9% | 16/104, 15.4% | 2.10 (0.65–6.84) | .217 |
| Non-White patients | ||||
| Outcome | Intensive treatment (n=369) | Standard treatment (n=370) | Odds ratio a (95% CI) for intensive treatment | p value |
| Good outcome | 240/369, 65.0% | 239/371, 64.4% | 1.15 (0.78–1.69) | .490 |
| Death | 13/369, 3.5% | 13/371, 3.5% | 0.93 (036–2.42) | .888 |
| Hematoma expansion | 48/366, 13.1% | 65/362, 18.0% | 0.61 (0.40–0.93) | .022 |
| Neurologic deterioration within 24 h | 36/369, 9.8% | 25/371, 6.7% | 1.36 (0.73–2.54) | .329 |
| Serious adverse event within 3 months | 74/369, 20.1% | 60/371, 16.2% | 1.16 (0.73–1.84) | .528 |
| Withdrawal of care | 6/369, 1.6% | 10/371, 2.7% | 0.45 (0.13–1.62) | .223 |
aAdjusted for patient age, intubation, baseline NIH Stroke Scale, and baseline intraparenchymal and intraventricular hemorrhage volume.
Discussion
In this post hoc analysis of the ATACH trial, we found race-related differences in the functional outcome after ICH and in response to BP lowering treatment. In adjusted models, Black patients had the lowest probability of good outcome compared to other race/ethnic groups. When we examined the treatment*race interaction, we found that White patients who received intensive BP treatment had lower predicted probability of good outcome compared to standard treatment. Otherwise, there were no differences in the functional outcome between both treatment arms in other groups.
There are race/ethnic-related disparities in presentation, risk factors, and outcome after ICH.11–15 Analysis of the Virtual International Stroke Trials Archive and Efficacy of Nitric Oxide in Stroke trial demonstrated no difference in 90-day mRS between Black, Asian, and White patients. The results were echoed in a subsequent analysis of the Ethnic/Racial Variations of Intracerebral Hemorrhage study which also demonstrated no race/ethnic differences in 3-month functional outcome after ICH. 14 However, an analysis of the Get With The Guidelines-Stroke (GWTG-Stroke) registry found that Black race was associated with lower odds of good outcome after ICH which is in accordance with our findings. 15 The inconsistency in these results may be attributed to differences in study design and inclusion criteria. For example, previous studies included both supratentorial and infratentorial ICH, unlike the present study which only included supratentorial hemorrhage.
Interestingly, the interaction between treatment and race/ethnicity was only significant in White patients, who had lower probability of good outcome when receiving intensive BP reduction. In addition, intensive treatment was associated with a higher risk of serious adverse events in White patients but not in non-White patients. Surprisingly intensive treatment was also associated with a higher risk of hematoma expansion in White patients and a lower risk of hematoma expansion in non-White patients. The reason for the difference in response to BP lowering treatment between White and non-White is not clear and requires further investigation. It is possible that the differences in hemorrhagic stroke etiology, risk factors, and response to antihypertensive medications between White and non-White patients mediated the difference in response to BP lowering treatment. For example, White patients in our cohort were older, and more likely to have atrial fibrillation and cardiac ischemic disease which could have potentially made them more prone to serious adverse events with BP lowering treatment.
Future studies are needed to understand the differences in BP trajectories, autoregulation physiology, and response to BP lowering treatment after acute stroke among different race/ethnicity groups. Moreover, our study highlights that Black patient were underrepresented in trials that investigated BP management strategies after intracerebral hemorrhage and future studies focused on Black patients may be warranted.
Limitations
Despite the large sample size and the prospective nature of the study, our study has multiple limitations that need to be taken into account. First, our analysis was exploratory analyses and prone to chance findings. Second, despite the large sample size of the overall study cohort, the sample size of each race/ethnic group is modest. Finally, our results cannot be applied to patients with infratentorial hemorrhage.
Conclusion
In the ATACH-2, Black patients had a lower probability of good outcome in adjusted models. In addition, there was an interaction between race/ethnicity and response to BP lowering treatment after ICH, with White patients having an association between intensive blood pressure reduction and worse outcome.
Footnotes
Declaration of conflicting interests: The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr de Havenon reports investigator-initiated funding from AMAG and Regeneron pharmaceuticals.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr de Havenon is supported by NIH-NINDS K23NS105924.
Ethical approval: The anonymized ATACH-2 dataset, received with a local IRB waiver
Informed consent: All participants or their legally authorized representative provided written informed consent before randomization
Trial Registration: ClinicalTrials.gov number, NCT01176565
ORCID iDs
Mohammad Anadani https://orcid.org/0000-0002-7813-2949
Adam de Havenon https://orcid.org/0000-0001-8178-8597
References
- 1.Moullaali TJ, Wang X, Martin RH, et al. Blood pressure control and clinical outcomes in acute intracerebral haemorrhage: A Preplanned Pooled Analysis of Individual Participant Data. Lancet Neurol 2019; 18: 857–864. [DOI] [PubMed] [Google Scholar]
- 2.Qureshi AI, Palesch YY, Barsan WG, et al. Intensive blood-pressure lowering in patients with acute cerebral hemorrhage. New Engl J Med 2016; 375: 1033–1043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Tsivgoulis G, Katsanos AH, Butcher KS, et al. Intensive blood pressure reduction in acute intracerebral hemorrhage: A Meta-Analysis. Neurology 2014; 83(17): 1523–1529. [DOI] [PubMed] [Google Scholar]
- 4.Sandset EC, Anderson CS, Bath PM, et al. European Stroke Organisation (ESO) guidelines on blood pressure management in acute ischaemic stroke and intracerebral haemorrhage. Eur Stroke J 2021; 6(2): I-XLVII. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
- 5.Gupta AK. Racial differences in response to antihypertensive therapy: does one size fits all?. Int J Prev Med 2010; 1: 217. [PMC free article] [PubMed] [Google Scholar]
- 6.Jones DW, Hall JE. Racial and ethnic differences in blood pressure: biology and sociology. Circulation 2006; 114(25):2757–2759. [DOI] [PubMed] [Google Scholar]
- 7.Profant J, Dimsdale JE. Race and diurnal blood pressure patterns: a review and meta-analysis. Hypertension 1999; 33: 1099–1104. [DOI] [PubMed] [Google Scholar]
- 8.Wang X, Poole JC, Treiber FA, et al. Ethnic and gender differences in ambulatory blood pressure trajectories. Circulation 2006; 114: 2780–2787. [DOI] [PubMed] [Google Scholar]
- 9.Gupta AK, Poulter NR, Dobson J, et al. Ethnic differences in blood pressure response to first and second-line antihypertensive therapies in patients randomized in the ASCOT Trial. Am J Hypertens 2010; 23: 1023–1030. [DOI] [PubMed] [Google Scholar]
- 10.Tibshirani R. Regression shrinkage and selection via the lasso. J R Stat Soc Ser B (Methodological). 1996; 58(1): 267–288. [Google Scholar]
- 11.Flaherty ML, Woo D, Haverbusch M, et al. Racial variations in location and risk of intracerebral hemorrhage. Stroke 2005; 36: 934–937. [DOI] [PubMed] [Google Scholar]
- 12.Koch S, Elkind MS, Testai FD, et al. Racial-ethnic disparities in acute blood pressure after intracerebral hemorrhage. Neurology 2016; 87: 786–791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Krishnan K, Beishon L, Berge E, et al. Relationship between race and outcome in asian, black, and Caucasian patients with spontaneous intracerebral hemorrhage: Data from the Virtual International Stroke Trials Archive and Efficacy of Nitric Oxide in Stroke trial. Int. J Stroke 2018; 13: 362–373. [DOI] [PubMed] [Google Scholar]
- 14.Miyares LC, Falcone GJ, Leasure A, et al. Race/ethnicity influences outcomes in young adults with supratentorial intracerebral hemorrhage. Neurology 2020; 94: e1271–e1280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Xian Y, Holloway RG, Smith EE, et al. Racial/ethnic differences in process of care and outcomes among patients hospitalized with intracerebral hemorrhage. Stroke 2014; 45: 3243–3250. [DOI] [PubMed] [Google Scholar]