Leukoaraiosis and intracerebral hemorrhage after thrombolysis in acute stroke

The term leukoaraiosis describes an abnormal CT appearance of the subcortical brain white matter, seen usually as bilateral (patchy or diffuse) areas of hypoattenuation in the periventricular regions or extended to the centrum semiovale.¹ As shown by studies with magnetic resonance perfusion,² these changes are likely caused by chronic ischemia (either diffuse and subliminal or focal and repeated) in the distal deep arterial or arteriolar territories. Lipohyalinosis (arteriolosclerosis), induced by chronic hypertension or diabetes or both, is the most common finding in the small vessels of deep brain from patients with white matter disease. Leukoaraiosis is a documented risk factor for ICH during oral anticoagulation treatment given for secondary prophylaxis after a cerebral ischemic event.³ Although its role in tissue plasminogen activator (tPA)–related hemorrhages is uncertain, there is a widespread concept among treating physicians that patients with extensive leukoaraiosis on baseline CT scan are more likely to develop symptomatic bleeding after tPA administration. In the present study, we sought to explore the relationship between leukoaraiosis and tPA-related hemorrhage by systematically reviewing the CT scans of a large prospective registry of patients with acute stroke treated with IV tPA across Canada: the Canadian Alteplase for Stroke Effectiveness Study (CASES).⁴

Methods.

Detailed methods and results from the CASES study have been published previously.⁵ The CASES Study is a prospective observational cohort study assessing the safety and effectiveness of IV tPA for acute ischemic stroke. The study was mandated by the federal government as a condition of licensure of tPA for the treatment of acute stroke in Canada. Over 2.5 years (from 1999 to 2001), a total of 1,135 consecutive patients were enrolled at 60 centers across Canada. Each center obtained institutional ethical approval for the data collection protocol. Outcomes were collected at 90 days after stroke and rated using the modified Rankin Scale (mRS) to measure functional dependence. Symptomatic ICH within 24 hours from thrombolytic treatment was defined as any neurologic deterioration judged by the treating physician to be secondary to a new brain hemorrhage as shown in a head CT or MRI. Copies of head CT scans obtained at baseline and at 24 hours after treatment were submitted to a central panel for review. We explored the relationship between leukoaraiosis and risk of symptomatic ICH by retrospectively reviewing the baseline CT scans of the CASES cohort for the presence of leukoaraiosis, old lacunar infarcts, and old territorial infarcts. Pretreatment CT scans were assessed independently and blindly to both clinical data and follow-up scans by two readers (V.P., J.M.B.). An interobserver reliability test in 40 scans was preliminarily performed between the two readers. The presence and extent of leukoaraiosis was assessed using Van Swieten scale.⁶ The Van Swieten scale is a simple scale rating separately the anterior and posterior region of the axial brain CT scan on a three-point scale from 0 (no white matter hypodensity) to 2 (confluent white matter hypodensity from the ventricles to the gray matter). The total score, obtained by summing the two scores in the anterior and posterior region, can vary from 0 to 4. In the present article, we separately assessed each hemisphere (rating separately the anterior and posterior region within each hemisphere) for a total score of 0 to 8 when considering the whole brain. Based on the distribution of the scores, we dichotomized Van Swieten score, with a total score >4 (the 10th percentile) denoting severe white matter disease. The presence and number of old lacunes (defined as cavitary lesions 15 mm or less in maximum diameter located in corona radiata, basal ganglia, or internal capsule regions) and territorial infarcts were recorded. We defined multilacunar state in the presence of two or more lacunes (fifth percentile) on baseline CT scan.

Outcome at 90 days was assessed using the modified Rankin Scale (mRS), dichotomized as excellent outcome (mRS 0 to 1) and disability/death (mRS 2 to 6).

Results.

Of the 1,135 patients enrolled in the trial, 936 baseline CT scans were available for review in the CASES main study; 116 were missed in the present study, so that 820 baseline CT scans were available and assessable for the presence of white matter changes and lacunes.

The interobserver reliability between the two readers showed near perfect agreement, with kappa values ranging from 0.9 to 1.0 for the grading of white matter disease in the four regions in examination (left and right anterior white matter, left and right posterior white matter), and a kappa value of 0.8 for the record of number of lacunes.

Seventy-one (8.6%) patients had evidence of severe white matter disease (Van Swieten score > 4) and 70 (8.5%) had two or more lacunar infarcts. As expected, these patients were significantly older than those with no/moderate white matter disease, and were more likely to have a history of hypertension and dementia (table 1).

From the univariable analysis, 8.4% of the patients with extensive white matter disease (Van Swieten score > 4) developed symptomatic ICH as a complication of thrombolytic treatment, compared to 3% of patients with no or moderate leukoaraiosis (RR 2.75; 95% CI 1.15 to 6.53). Similarly, symptomatic ICH occurred in 10% of patients with multiple lacunes, and in 2.9% of patients with one lacune or less (RR 3.40; 95% CI 1.50 to 7.68) (table 2). After multivariable adjustment for baseline characteristics and other possible predictors of symptomatic ICH (age, blood glucose, ASPECTS, baseline NIHSS, time to treatment, blood pressure, and other baseline characteristics), we found no substantial differences from the univariable risk ratios. Only glucose was an independent predictor of symptomatic ICH⁷; however, glucose did not confound the relationship between symptomatic ICH and Van Swieten score or lacunar state. Similarly, there was no evidence of a glucose by lacunar state or glucose by Van Swieten score interaction in predicting symptomatic ICH.

There was no relationship between rate of symptomatic ICH and number of cortical territorial infarcts.

At 3 months outcome, univariable analysis showed that both two or more lacunes and a Van Swieten score >4 were associated with poor outcome (mRS ≥ 2). However, this association was confounded principally by age and baseline stroke severity, since this effect was not maintained after multivariable analysis. Death at 90 days was associated with more extensive white matter disease (higher van Swieten score) in univariable analysis, but this effect turned out to depend on age and baseline NIHSS score after adjustment for baseline characteristics. Conversely, death at 90 days was more frequent among patients with two or more lacunes, an association that persisted after adjustment for age, baseline NIHSS score, baseline serum glucose, and baseline ASPECTS score (OR 2.9; 95% CI 1.3 to 6.2, p = 0.008) (figure).

• Download figure
• Open in new tab
• Download powerpoint

Figure. Ninety days outcome, stratified by number of lacunes, in the Canadian Alteplase for Stroke Effectiveness Study cohort.

No interaction between age and van Swieten score was observed to suggest a differential predictive role for white matter at any age: this was true for all outcomes measures (death, excellent outcome, and symptomatic ICH).

Discussion.

The results from the present study show that the presence of severe small vessels disease, revealed by extensive leukoaraiosis and lacunar infarcts on neuroimaging, increases the risk of cerebral bleeding complications after tPA administration. In our study 8.4% of patients with leukoaraiosis and 10% of patients with multiple lacunes on baseline CT developed symptomatic ICH after systemic thrombolysis. Although patients with severe leukoaraiosis were significantly older compared to the group with no or moderate leukoaraiosis, and were more likely to have a history of hypertension and dementia, the multivariable analysis showed that the increase of bleeding risk determined by the presence of leukoaraiosis or multiple lacunes was independent by age and any other factor reported to increase this risk. Our data confirm the results of a recent MRI-based retrospective study in 449 patients with acute ischemic stroke treated with thrombolysis within 6 hours from symptoms onset, showing a symptomatic hemorrhagic rate of 10.5% in the presence of moderate to severe leukoaraiosis⁸: the observation of an almost exact correspondence between the two studies supports our results, with the advantage that CT is widely available and commonly used to evaluate patients with acute stroke in the majority of centers.

The correlation between leukoaraiosis and ICH, first reported in 1989,⁹ has been shown to be almost fully explained by the higher prevalence of arterial hypertension and lacunar infarcts on CT scan.¹⁰ Among patients with leukoaraiosis, hemorrhage was more frequently located in the deep basal ganglia, confirming the hypothesis that the association was due to the common underlying pathology, i.e., small-artery disease on a hypertensive basis. In recent years, gradient-echo T2*-weighted MRI sequences, showing the presence of silent microbleeds, have corroborated this hypothesis: in one study, the number of deep microbleeds was correlated with the severity of periventricular hyperintensities on MRI and with the presence of lacunar infarcts.¹¹ The concomitant presence of microbleeds in patients with extensive white matter disease could explain the observed higher rate of symptomatic ICH in our study, in view of the correlation between imaging signs of cerebral microangiopathy, clinically silent microbleeds, and intraparenchymal hemorrhage.^12,13 To support this hypothesis, clinically silent microbleeds in the corticosubcortical area and deep gray matter have been shown to be strongly associated with hemorrhage in the same area.¹⁴ In a recent pooled analysis of 600 patients who had a pretreatment MRI scan, T2*-weighted images with evidence of microhemorrhages were not shown to be predictors of symptomatic ICH after thrombolytic treatment, although no data about the presence of leukoaraiosis were reported.¹⁵ Therefore, while this could be an attractive explanation for increased symptomatic ICH among treated patients, it remains unclear what the mechanism is.

The Stroke Prevention in Reversible Ischemia Trial (SPIRIT)¹⁶ pointed to a definite, prospectively assessed, independent role of leukoaraiosis as a risk factor for major bleeding during anticoagulation after cerebral ischemia: patients with leukoaraiosis had a risk of developing symptomatic ICH almost three times higher than patients without leukoaraiosis. Although the presence of high INR values in the SPIRIT trial (between 3.0 and 4.5) could have been a confounder, a case-control study³ aiming to investigate radiographic and clinical characteristic of patients with warfarin-related ICH after ischemic stroke (cases) compared with patients treated with warfarin and without ICH (controls) showed that the presence and severity of leukoaraiosis on CT scan was strongly correlated with the occurrence of ICH at any INR value, an association that persisted after controlling for other risk factors.

The CASES study also shows that a multilacunar state is a poor prognostic marker: death at 90 days was more common among patients with two or more lacunes and patients with radiologic hallmarks of small vessels disease had a poorer outcome (22.5% of patients with leukoaraiosis vs 37% with no or moderate leukoaraiosis had mRS 0 to 1 at 90 days). The 3-month outcome in patients with severe leukoaraiosis is consistent with previously published outcome data in patients older than 80 years.^17,18 In the CASES Study, 26% of elderly patients had excellent outcome at 90 days, with a mortality rate of 35%. Very old age has been demonstrated to be an independent predictor of outcome and mortality.¹⁹ In our study, mortality at 90 days appears to be higher, reaching 40% in patients with extensive white matter disease; however, combined dependent outcomes (mRS > 2) are not predicted by multiple lacunes or van Swieten score > 4. Taking into account that patients with extensive small vessels disease are significantly older than those with no or moderate white matter changes, it can be hypothesized that the presence of a multilacunar state may be an independent marker of severe cerebrovascular disease in older patients.

The major limitation of the present study was the lack of controls: our data do not result from a randomized trial so they should be interpreted with caution. Further, our data represent a subset of the entire CASES study due to missing scans. Although we have no reason to suspect it, if the missing scans were systematically different from those included, our results may be biased. Although strengthened by their consistency with recently published data from an MRI study, our arbitrary choice of cut-points for white matter hypodensity and lacunes may have biased the findings toward positive results. Nevertheless, these observational data may be used by clinicians in counseling patients and families about the expected risks of therapy.

Acknowledgment

The authors thank Dr. Leonardo Pantoni for his contribution to data interpretation and improvement of the final version of the manuscript.

Appendix

CASES Writing Committee: Chairs—Michael D. Hill, Alstair M. Buchan. Members—Stephen Phillips, Robert Cote, Richard Riopelle, Frank Silver, Daniel Selchen, Betty-Anne Schwarz, Vladimir Hachinski, Ashfaq Shuaib, Philip Teal.

The CASES Investigators: Alberta—Alastair M. Buchan (Co-Principal Investigator), Michael D. Hill (Co-Principal Investigator), Andrew M. Demchuk, Keith M. Hoyte, Gary M. Klein, Philip A. Barber, Nancy Newcommon, Andrea Cole-Haskayne, Zeenie Ramji, Diana Czechowsky, William Ghali, Rollin Brant, J.H. Warwick Pexman, Ashfaq Shuaib, Edina Kadribasic, Toni Winder, Carolyn Walker, Prafull Parekh, Scott Wilson, Yolande Westra, Robert Burris, Gail Deagle, Laurel Reynolds. British Columbia—Philip Teal, Andrew Woolfenden, Claire Johnston, Todd Collier, Judy Moa, Donald Cameron, Loree Tadey, Cate Earl, Tracy Sacre, Kennely Ho, Frank Kemble, Elizabeth Mauthe, Lynda MacFarlane, David Novak, Richard Grosch, Joe Haegart, Cathy Metcalfe, Dr. Milton Wong, Barbara Boychuk, Terry Curran, Suzanne Bailey, Anthony Costantino, Helen Costantino, David Craig, Michelle Mantle, Judith Huff, Michael Kenyon, Cutler, Chuck Mahoney, Jo Thoburn. Manitoba—Brian A. Anderson, Mohammad Nagaria, Dan F. Gladish, Brian Schmidt, Douglas Eggertson, Douglas Hobson, Lawrence Hudson, Greg McGinn, Farid Esfahani. New Brunswick—Peter Bailey, Gregg MacLean, A. MacDougall, Peggy Cook, Shelly Alward, Mary Freeman-McCraek, Lynn Reid, Byrne Harper. NFLD—Alan Gooderidge, William Pryse-Phillips, Mark Stefanelli, Jim Scott. Nova Scotia—Stephen Phillips, Gordon Gubitz, Judith Jarrett, Richard Leckey, G. Maharaj. Ontario—Rudolph Arts, Sylvia Styling, Vladimir Hachinski, J. David Spence, Brian Silver, Bart M. Demaerschalk, Ashok Devasenapathy, Jose G. Merino, Edward H. Wong, J. Arturo Tamayo Mendoza, Blaine Foell, Fali Poncha, Connie A. Frank, Mary M. McTaggart, Frank Silver, Cheryl Jaigobin, Shelley Yantha, Pamela Urzua, Relu Wiegner, Bob Duke, Wes Oczkowski, David Fletcher, Sera Nicosia, Nancy Pyette, Sandy Black, Na Jiang, David Gladstone, Edwin Klimek, Chidambaram Yegappan, Marianne Klimek, Jeanine Robinson, Daniel Selchen, Alnoor Dhanani, Bryan Temple, Heather Hink, Deidre Davidson, Kathryn Leblanc, Suzanne Christie, Lucian Sitwell, Andre Douen, Antoine Hakim, Nicole Pageau, Hyman Rabinovitch, Martin Del Campo, Neville Bayer, David Morgenthau, Gabrielle De Veber, Andrew Kertesz, Darlyne Morlog, Richard Magder, Garry Moddel, Chidambaram Yegappan, Iram Zando, Martin Lees, Peter Stys, Betty Anne Schwarz, David Howse, Richard Riopelle, Cindy Bolton, Sandy Weatherby, Dianne Groll, Mukul Sharma, Hiren Desai, Michael Winger, Dave Tamblyn, Stephane Sauve, Hui Lee, Deon Leow, Dwight Stewart, Susan Fawcett, Vince DePaul. Quebec—Robert Cote, Liam Durcan, Anne-Marie Fontaine, Lisa Wadup, Francois Jacques, Denis Halle, André Bellavance, Leo Berger, Linda Moisan, Jeanne Teitelbaum, Nicole Lachance, Ariane Mackey, Denis Simard, Annette Hache, Barbara Léger, Lison Fournier, Sophie Dubé, Jeffrey Minuk, Louise-Hélène Lebrun, Andre Durocher, Nicole Daneault, Sylvain Lanthier, Marlene Lapierre, Marie-Paul Desrochers, Michel Beaudry, Doris Boivin, Donald Rivest, Marc Petitclerc, Marie Berberl, Charles Beaurivage, Albert Lamontagne, Serge Desviens. Saskatchewan—Ali Rajput, Christopher Voll, Michele Rajput, Aileen Schultz, Felix Veloso, Barbara Seal, L. Moodley.

CASES CT Review Group: J.H. Warwick Pexman, Philip A. Barber, Andrew M. Demchuk, Michael D. Hill.

Hoffmann-La Roche Representatives: Alberta—Alanna Campbell, Frank Heron. British Columbia—Kath Davy. Manitoba/Saskatchewan—Kath Davy. Ontario—Angie McAllister. Quebec—Rene Garneau. Atlantic Canada—Linda Mosher. Administrative Asst.: Vivian Thompson. Regional Business Manager: Jean-Rene Poirier. Associate Medical Directors: Kate Lawrence, Esther Tan. Director of Marketing, Hospital Care: Alan LeVoguer.

Canadian Stroke Network: President—Antoine Hakim. Heart & Stroke Foundation of Canada: Elinor Wilson. Canadian Stroke Consortium: President—Philip Teal. Past President—John Norris. Staff—Donna Huber.