Treatment of acute ischemic stroke in patients with cerebral microbleeds: a decision analysis

Treatment with anti-thrombotic agents reduces the likelihood of recurrent ischemic stroke. However, it may produce intracerebral hemorrhage. The risk of recurrence is highest shortly after an ischemic stroke has occurred irrespective of the ischemic stroke subtype. About one-fourth1 to one-half2 of the risk of recurrence at 1 year is accrued during the acute phase (first 2–4 weeks) of an ischemic stroke, and this risk declines further during the following 5 years.3 Similarly, the rates of intracerebral hemorrhage decline from 0.6% to 3% during the the first 2–4 weeks after an ischemic stroke4–13 to 0.16–0.23% annually thereafter.11,14 This time variance in complication rates indicates that the risk–benefit ratio of the treatment of patients with acute ischemic stroke may be different from that of their long-term care. Still, published analyses of the treatment of such patients have assumed that the risks of ischemic stroke recurrence and intracerebral hemorrhage are constant over the patient’s lifetime,15 or over the first year after ischemic stroke.16 Therefore, the conclusions of these analyses may not pertain to patients with acute ischemic stroke. The management of ischemic stroke in patients with cerebral microbleeds is particularly uncertain. About one-fourth of the patients with ischemic stroke have cerebral microbleeds that may predict a higher risk of intracerebral hemorrhages.17 Still, we know of no randomized studies comparing the risks and benefits of anti-thrombotic treatment in patients with and without cerebral microbleeds. When clinical decisions must be made in the absence of evidence, an analysis of the benefits (efficacy) and risks (adverse effects) of alternative management options may identify the treatment of choice and the specific areas of uncertainty that should be explored by future research. In this article, we use a decision tree to compare no treatment, aspirin or anti-coagulation in patients with acute ischemic stroke who are not eligible for thrombolysis. Specifically, an attempt is made to (i) analyze the risks and benefits of these treatments; (ii) explore whether a finding of cerebral microbleeds affects the risk–benefit ratio of anti-thrombotic treatment; and (iii) identify the sources of uncertainty in treating patients with acute ischemic stroke with and without cerebral microbleeds. We used Paper Chase18 to search Medline until February 2010 for cohort studies of patients with ischemic stroke, from which we could derive the hazards of recurrent ischemic stroke and intracerebral hemorrhage, and the effect of treatment and cerebral microbleeds on these hazards. The term ‘hazard’ refers to the probability of a subject developing an event (a fatal or nonfatal recurrent ischemic stroke or intracerebral hemorrhage) during a discrete time interval (up to 30 days after an acute ischemic stroke). For studies of recurrent ischemic stroke, we used the terms: (‘cohort studies’ or ‘follow up studies’) and (‘ischemic attack, transient’ or ‘cerebral infarction’) and (‘recurrence’). We found 11 studies of patients aged ≥65 years that had (i) stratified the hazards of recurrent ischemic stroke by the subtype of the index ischemic stroke (large vessel, lacunar, cardio-embolic and cryptogenic) and (ii) provided either recurrence rates within 30 days, or Kaplan–Meier graphs, from which these rates could be derived by measuring the decline in the percentage of stroke-free survival during the acute phase of ischemic stroke (Table 1). To estimate the risk of intracerebral hemorrhage, we used the terms (‘cerebral hemorrhage’) and (‘fatal outcome’ or ‘platelet aggregation inhibitors’ or ‘anti-coagulants’). To assess the risk of ischemic stroke and intracerebral hemorrhage in patients with cerebral microbleeds, we used the terms (‘cerebral infarction’ or ‘cerebral hemorrhage’) and (‘microbleeds’ or ‘cerebral amyloid angiopathy’). Table 2 lists the baseline values for the efficacy and adverse effects of the treatment of patients with acute ischemic stroke. The sources of these values are presented in the following sections. Reported hazards of recurrent cerebrovascular ischemic events in patients with an average age of >65 years by subtype of the index ischemic stroke aValues approximated from Kaplan–Meier graphs. Reported hazards of recurrent cerebrovascular ischemic events in patients with an average age of >65 years by subtype of the index ischemic stroke aValues approximated from Kaplan–Meier graphs. Baseline values and ranges of published data on the disease- and treatment-related risks that were used in the analysis of the choice of treatment of acute ischemic stroke in elderly patients with and without cerebral microbleeds RR: relative risk; CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. Baseline values and ranges of published data on the disease- and treatment-related risks that were used in the analysis of the choice of treatment of acute ischemic stroke in elderly patients with and without cerebral microbleeds RR: relative risk; CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. We used TreeAge (TreeAge Software, Inc, 2008) to examine the choice between no treatment, aspirin or anti-coagulation (heparin or low-molecular weight heparinoids followed by oral anti-coagulants) in patients with or without cerebral microbleeds during the acute phase (2–4 weeks) after ischemic stroke. With any of these options, patients may either remain well or sustain a recurrent ischemic stroke and intracerebral hemorrhage. Those who sustain one or both of these may either survive or die. Incidence rates of, and case fatalities after recurrent ischemic stroke and intracerebral hemorrhage are presented as percentages. Probabilities are derived from these rates and presented on a 0–100 scale using the equation: Probability P = 1 − exp(−rates). The outcomes of the analysis are presented as probabilities of (i) developing a recurrent ischemic stroke and intracerebral hemorrhage and (ii) dying after a recurrent ischemic stroke and intracerebral hemorrhage. To identify sources of uncertainty, we determined the sensitivity of these outcomes to variations in each variable within the range of published data. The phrase ‘sources of uncertainty’ refers to those variables in which the thresholds between alternative management options (i.e. the values beyond which the optimal treatment changes) were within this range. We assumed that the various types of anti-coagulation (heparin and oral anti-coagulants) carry the same risk of bleeding. We biased the decision tree against ‘no treatment’ by adopting a high-bound estimate of the mortality rates after recurrent ischemic stroke. We also biased the decision tree in favor of anti-coagulation by (i) adopting low-bound estimates of intracerebral hemorrhage rates and intracerebral hemorrhage case fatality, and of the effect of cerebral microbleeds on these risks; (ii) considering only intracerebral hemorrhage, but not other intracranial and extracranial hemorrhages; and (iii) assuming that anti-coagulants are more efficacious than aspirin in reducing recurrence rates during the first 30 days after ischemic strokes, even though the advantage of anti-coagulation has been documented only for the long-term prevention of cardio-embolic strokes.19 After excluding the outlayers, reported recurrence rates during the first 30 days after an acute ischemic stroke have varied between 2.4% and 18.5% after large vessel strokes, 0–5.5% after lacunar strokes, 3.5–7% after cardio-embolic strokes and 1.5–6% after cryptogenic strokes (Table 1). The 30-day case fatality after ischemic stroke has been reported to be 5–17% after a first ever ischemic stroke49–51 and 23% after a recurrent ischemic stroke.3 We assumed baseline recurrence rates of 10% for large vessel, 2% for lacunar, 5% for cardio-embolic and 4% for cryptogenic strokes, and a high-bound 23% case-fatality rate after recurrent ischemic stroke (range: 5–23%). The phrase ‘baseline ischemic stroke recurrence rates’ refers to the proportion of patients without cerebral microbleeds, who would be expected to sustain a recurrent ischemic stroke if left untreated. Meta-analyses of randomized trials have indicated that aspirin reduced ischemic stroke recurrence rates by 30% (relative risk, RR = 0.7) during the first three weeks8,14 and 20–23% thereafter,14,52 and that heparin treatment, started within 48 h after IS, reduced recurrence rates by 20% (RR = 0.8)36 and 40% (RR = 0.6).37 A meta-analysis of long-term trials indicated that, in patients with atrial fibrillation, warfarin reduced the risk of ischemic stroke by 60% (RR = 0.4).19 We assumed that the RR of recurrent ischemic stroke with vs. without anti-coagulant treatment was 0.4 for patients with acute cardio-embolic ischemic stroke, and 0.6 (range: 0.6–0.8) for patients with the remaining ischemic stroke subtypes. In patients with acute ischemic stroke, the term ‘intracerebral hemorrhage’ refers to either hemorrhagic transformations of the infarction or ‘extra-ischemic’ hematomas (remote from the infarct). To detect hemorrhagic transformations, authors have variably used computed tomography (CT);6,9,12,53 CT or autopsy;7 and CT or magnetic resonance imaging (MRI).38,39 Some authors have differentiated between hemorrhagic transformations—types 1 and 2 (small and confluent petechiae, respectively), and parenchymal hematomas—types 1 and 2 (less and more than 30% of the infarcted area, respectively), and concluded that only type 2 hematomas were clinically significant.6,53 Others have distinguished between symptomatic and asymptomatic hemorrhagic transformations; however, the criteria for defining a hemorrhagic transformation as ‘symptomatic’ were either not reported,5,9,39,53 or variably described as, ‘any CT-documented hemorrhage that was temporally related to deterioration in the patient’s clinical condition in the judgment of the investigator’,54 or ‘an increase of 4 or more points in the NIHSS score’.55 This diversity in definitions and methods of detection has probably contributed to the variability in reported rates of intracerebral hemorrhage during the acute phase of ischemic stroke. Reported rates varied between 3% and 40% for ‘any hemorrhagic transformation,6,7,9,12,53 0.6–3% for ‘symptomatic hemorrhagic transformations’,4,5,8,10,11,53 and 1–2% for ‘parenchymal hematoma type 2’.6,9,11,13 Two studies of 10038 and 4139 patients differentiated between hemorrhagic transformations and extra-ischemic hematomas in patients with acute ischemic stroke, and they both found that 2% of the patients had hematomas remote from the infracted area on admission39 or within 7 days.38 Three studies stratified the risk of intracerebral hemorrhage during the first 7 days after an ischemic stroke by its subtype. They found that, after a lacunar stroke, this risk was nil12,56 or 0.1%.25 One study found a ‘parenchymal hemorrhage’ in 1.1% of the patients with large vessel, 2.7% with cardio-embolic and 2.6% with cryptogenic strokes.25 The second found a ‘hemorrhagic infarct’ in 26, 14 and 9%, respectively,56 and the third one found a ‘hemorrhagic transformation’ in 4, 29 and 40%, respectively.12 We derived our estimates from the reported rates of parenchymal hematomas type 2, extra-ischemic hematomas and symptomatic hemorrhagic transformations and assumed that the baseline risk of intracerebral hemorrhage was 0.1% for patients with lacunar strokes, and 1% (range: 0.4–2%) for patients with the remaining ischemic stroke subtypes. The term ‘baseline intracerebral hemorrhage risk’ refers to the proportion of patients who would be expected to sustain an intracerebral hemorrhage in the absence of any anti-thrombotic treatment. Meta-analyses of randomized trials have indicated that RRs of intracerebral hemorrhage in aspirin treated vs. untreated patients with ischemic stroke were 1.2,11 1.38 and 2.1.14 The incidence of intracerebral hemorrhage in heparin treated vs. untreated patients with acute ischemic stroke has been reported to be 2.6% vs. 1.1% (RR = 2.4),43 1.4% vs. 0.5% (RR = 2.8)36 and 2.5% vs. 0.7% (RR = 3.6).37 We adopted RRs of 1.7 (range 1.2–2.1) and 3.0 (range: 2.0–4.0) for aspirin and anti-coagulation, respectively. A population-based study found single cerebral microbleeds in 18% at age 60–69 years, and 38% at age 80–89 years and multiple cerebral microbleeds in 5% at age 60–69 years and 23% >80 years.41 Cerebral microbleeds appear to be associated with both ischemic stroke and intracerebral hemorrhage. A recent meta-analysis found that cerebral microbleeds occurred in 5% of elderly subjects without cerebrovascular disease; 23% of the patients with first-ever ischemic stroke; 44% with recurrent ischemic stroke; 52% with first-ever intracerebral hemorrhage; and 83% with recurrent intracerebral hemorrhage.17 The association between cerebral microbleeds and intracerebral hemorrhage is supported by several observations. First, cerebral microbleeds are regionally associated with intracerebral hemorrhage57 and with large-sized intracerebral hemorrhages.58 Second, patients with ischemic stroke and cerebral microbleeds were 3–24 times more likely to sustain an intracerebral hemorrhage during follow-up periods of 14–33 months than those without cerebral microbleeds.32–35,41,42 However, the association between cerebral microbleeds and ischemic stroke is uncertain. It has been reported that cerebral microbleeds were associated only with lacunar infarcts, but not cortical infarcts,59 and longitudinal studies have found that the RR of a recurrent ischemic stroke in patients with and without cerebral microbleeds varied between 1 (no added risk)32,34 and 1.3–2.3.33,35 The relationship between cerebral microbleeds and intracerebral hemorrhages during the acute phase of ischemic stroke is similarly uncertain. Two studies found no differences in the risk of hemorrhagic transformations in patients with vs. without cerebral microbleeds.39,60 On the other hand, it has been reported that new cerebral microbleeds developed within 1 week after acute ischemic stroke in 13% of the patients, mostly in those with past cerebral microbleeds.61 Furthermore, a regression analysis indicated that cerebral microbleeds were independent predictors of both hemorrhagic transformations and hematomas remote from the infarcted area, with an adjusted odds ratio of 7.2.38 We estimated that patients with acute ischemic stroke and cerebral microbleeds were three times more likely to sustain an intracerebral hemorrhage, and 1.5 times more likely to sustain a recurrent ischemic stroke than patients without cerebral microbleeds. However, considering the variability in reported data, we used for the sensitivity analyses an RR range between 1 (i.e. no added risk in patients with cerebral microbleeds) and 7 for intracerebral hemorrhage, and 1–2.3 for recurrent ischemic stroke. The effect of cerebral microbleeds on the risk of intracerebral hemorrhage during anti-thrombotic treatment is uncertain. On the one hand, cerebral microbleeds are not believed to increase the risk of intracerebral hemorrhage after thrombolytic treatment.62,63 On the other hand, cerebral microbleeds47 and cerebral amyloid angiopathy64 have been reported to increase the risk for intracerebral hemorrhage in patients treated with aspirin or anti-coagulants, and regression analyses have indicated that cerebral microbleeds were independently associated with intracerebral hemorrhage in such patients.35,65 Therefore, we assumed an additive effect of cerebral microbleeds and anti-thrombotic treatment on the risk of intracerebral hemorrhage. The case fatality after intracerebral hemorrhage in intensive care units has varied between 25% and 64%,40 and we adopted a baseline value of 30% (range: 25–64%). Regression analyses have indicated that, after adjusting for age and co-morbidity, anti-coagulants increased the case fatality after intracerebral hemorrhage, with odds ratios of 1.5,44 1.6,47 2.048 and 3.245, while aspirin increased the case fatality after intracerebral hemorrhage with odd ratios of 1.1,44 2.545 and 2.9.46 We assumed that the RRs of the 30-day case fatality are a low-bound value of 1.1 (range: 1.0–2.5) for aspirin and 1.5 (range: 1.5–3.5) for anti-coagulants. In patients without cerebral microbleeds, aspirin emerged as the treatment of choice for any ischemic stroke subtype (Table 3). Anti-coagulation was more efficacious than aspirin in preventing ischemic stroke recurrence; however, this gain was offset by the higher intracerebral hemorrhage rates. Probabilities of the outcomes of the analysis of the choice between no treatment, aspirin or anti-coagulants in elderly patients without cerebral microbleeds during the acute phase of an ischemic stroke IS: ischemic stroke; ICH: intracerebral hemorrhage. Probabilities of the outcomes of the analysis of the choice between no treatment, aspirin or anti-coagulants in elderly patients without cerebral microbleeds during the acute phase of an ischemic stroke IS: ischemic stroke; ICH: intracerebral hemorrhage. The choice of aspirin was robust over variations within the range for published data in most variables in the model. Aspirin remained better than anti-coagulation as long as the baseline intracerebral hemorrhage rates were assumed to exceed 0.3% in patient with large vessel ischemic stroke, 0.1% in those with lacunar IS, 0.2% in those with cryptogenic ischemic stroke and 0.4% in patients with cardio-embolic ischemic stroke; and as long as the RR of intracerebral hemorrhage in anti-coagulated patients was >1.9 in patients with large vessel ischemic stroke, 2.4 in those with lacunar ischemic stroke, 1.5 in those with cryptogenic ischemic stroke and 2.2 in patients with cardio-embolic ischemic stroke (data not shown). In patients with cerebral microbleeds, anti-coagulation was similarly associated with higher mortality rates than in patients treated with aspirin (Table 4). However, the choice between aspirin and no treatment was sensitive to variations within the reported range in the risk of intracerebral hemorrhage and ischemic stroke recurrence, in the case-fatality after intracerebral hemorrhage and recurrent ischemic stroke, in the increase in the risk of intracerebral hemorrhage following aspirin and in the effect of cerebral microbleeds on the risk of intracerebral hemorrhage and ischemic stroke recurrence (Table 5). Probabilities of the outcomes of the analysis of the choice between no treatment, aspirin or anti-coagulants in elderly patients with cerebral microbleeds during the acute phase of an ischemic stroke CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. Probabilities of the outcomes of the analysis of the choice between no treatment, aspirin or anti-coagulants in elderly patients with cerebral microbleeds during the acute phase of an ischemic stroke CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. Threshold values of the variables used in the analysis of the choice between aspirin and no treatment in elderly patients with cerebral microbleeds during the acute phase of an ischemic stroke RR: relative risk; CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. Threshold values of the variables used in the analysis of the choice between aspirin and no treatment in elderly patients with cerebral microbleeds during the acute phase of an ischemic stroke RR: relative risk; CMBs: cerebral microbleeds; IS: ischemic stroke; ICH: intracerebral hemorrhage. Three main findings emerge from the present analysis. First, anti-coagulation was associated with the highest probability of dying, even though our decision model was deliberately biased in favor of anti-coagulant treatment. Second, in patients without cerebral microbleeds, aspirin was the treatment of choice. This conclusion is consistent with the recommendation to treat patients with acute ischemic stroke, who are not eligible for thrombolysis, with aspirin, whatever the ischemic stroke subtype.66 Third, the analysis highlighted an important area of uncertainty: in patients with acute ischemic stroke and cerebral microbleeds, no treatment may be superior to aspirin. The question whether the presence of cerebral microbleeds reverses the risk–benefit assessment of anti-thrombotic treatment in patients with ischemic stroke has been discussed in the past.67 It has been suggested that, in patient with a high load of cerebral microbleeds, the risk of intracerebral hemorrhage may outweigh the benefit of aspirin.35 Our analysis confirms this uncertainty and identifies its main sources: the variability in reported intracerebral hemorrhagerates, and the conflicting data on the relationship between intracerebral hemorrhage, cerebral microbleeds and anti-thrombotic treatment. All of these sources of uncertainty are related to the diverse ways various authors define intracerebral hemorrhage, to the variability in the imaging techniques of cerebral microbleeds and to the lack of standardization of the number and location of the cerebral microbleeds that are thought to suggest a bleeding-prone microangiopathy.67 The effect of cerebral microbleeds on the outcomes of treatment is relevant for as many as one-fourth of the patients with acute ischemic stroke.19 Therefore, future studies should explore the sources of uncertainty that we identified in this article. This would require a rigorous definition of the terms intracerebral hemorrhage and cerebral microbleeds and of the methods of their detection. Hopefully, a standartization of the methods of imaging, and the number, size and location of cerebral microbleeds will help in elucidating their relationship with the outcomes of treatment of patients with acute ischemic stroke. Conflict of interest: None declared.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.