Article Text
Abstract
Objective To identify in patients who survived 6 months after a spontaneous intracerebral haemorrhage (ICH) baseline characteristics and new clinical events associated with functional decline.
Methods In a single-centre study, we prospectively included 6-month survivors with a modified Rankin Scale (mRS) score 0–3. We defined functional decline by a transition to mRS 4–5. We evaluated associations of baseline characteristics and new clinical events with functional decline, using univariate and multivariable models.
Results Of 560 patients, 174 (31%) had an mRS score 0–3 at 6 months. During a median follow-up of 9 years (IQR 8.1–9.5), 40 (23%) converted to mRS 4–5. Age, diabetes mellitus, ICH volume and higher mRS scores at 6 months were independently associated with functional decline. Among baseline MRI markers, presence of strictly lobar cerebral microbleeds (CMBs), and mixed lobar and deep CMBs were independently associated with functional decline. When new clinical events occurring during follow-up were added in multivariable models, age (cause-specific HR (CSHR): 1.07; 95% CI: 1.03 to 1.11), ICH volume (CSHR: 1.03; 95% CI: 1.01 to 1.06), mRS score at 6 months (CSHR per 1 point increase 1.61, 95% CI 1.07 to 2.43), occurrence of dementia (CSHR: 3.81, 95% CI: 1.78 to 8.16) and occurrence of any stroke (CSHR: 4.29, 95% CI: 1.80 to 10.22) remained independently associated with transition to mRS 4–5.
Interpretation Almost one-fourth of patients with spontaneous ICH developed a functional decline over time. Age, ICH volume, higher mRS score at 6 months and new clinical events after ICH are the major determinants.
Statistics from Altmetric.com
Introduction
Spontaneous intracerebral haemorrhage (ICH) is a devastating illness associated with a high short-term mortality rate that leads to severe disability among survivors.1 Previous studies have extensively evaluated short-term and 1-year functional outcomes.2–4 Little is known about the long-term functional decline, defined as a decrease of functional capacities. Only few population-based studies have addressed this topic. They suffered from methodological limitations such as small sample size and different timing of inclusion, while ICH management changed over time.5
Identification of determinants of functional decline would have a clinically meaningful impact on planning of care and preventive strategies to reduce the burden related to functional disability in ICH survivors. It may also inform future cost-effectiveness studies in ICH. Furthermore, as modified Rankin Scale (mRS) is used as the main endpoint in the majority of clinical trials evaluating therapeutic interventions in patients with ICH, determining temporal trajectories of functional decline over time might provide useful insight for the design of future trials.6
Baseline clinical and MRI characteristics associated with functional decline after an ICH are poorly known. Previous studies have suggested that arteriolosclerosis and cerebral amyloid angiopathy (CAA) are associated with different outcomes after ICH especially in terms of ICH recurrence and cognitive impairment.7–9 However, the influence of different types of vasculopathies on functional decline has never been tested. Finally, whether new clinical events after ICH are major determinants of functional decline is also not known.
In a prospective cohort of patients with spontaneous ICH and a favourable functional status at 6 months (mRS 0–3),4 we aimed at identifying baseline clinical and MRI characteristics associated with functional decline, and at assessing the influence of new clinical events (dementia, new ICH or ischaemic stroke) on functional decline.
Methods
Study design and participants
We included patients from the Prognosis of InTraCerebral Haemorrhage (PITCH) cohort, an observational and prospective study.7 10 We included consecutive patients older than 18 years (no upper age limit was imposed) admitted at the Lille University Hospital between 3 November 2004 and 29 March 2009 for a stroke.7 11 Patients with evidence of parenchymal haemorrhage on their first CT scan were included. Exclusion criteria were pure intraventricular haemorrhage, ICH resulting from intracranial vascular malformation, intracranial venous thrombosis, head trauma or tumour; and haemorrhagic transformation within an infarct. The present study focuses on functional decline occurring during follow-up. Therefore, we included patients who were still alive 6 months after ICH and had mRS scores 0–3.7 11
Demographic and clinical variables
We prospectively collected demographic characteristics, presence of vascular risk factors, history of previous stroke (either ischaemic or haemorrhagic), ischaemic heart disease, atrial fibrillation and ongoing treatments, using previously reported definitions.7 10
Imaging
All patients underwent a brain CT scan at admission. We classified ICH locations as previously defined and for our analyses we used only lobar (frontal, temporal, parietal, occipital when the origin appeared to be in the cerebral hemispheres superficial to the deep grey matter structures) and deep (when the haemorrhage originated from lenticular or caudate nucleus, thalamus, internal or external capsule, and brainstem).7 11 We calculated the volume of ICH according to the A×B×C/2 method.12
Patients without contraindications underwent a 1.5 Tesla brain MRI scan after a few days, when stable. The MRI protocol included at least fluid attenuated inversion recovery (FLAIR), and T2*-gradient-echo (GRE) weighted sequences (echo time: 22×8 ms, repetition time: 700 ms; flip angle: 25°; field of view: 250 mm; matrix: 352×224; slice thickness: 5 mm and interslice gap: 1.5 mm).7 The neuroradiological assessment was performed as extensively described in previous studies conducted in this cohort.8 We assessed white matter hyperintensities with the Fazekas Scale with scores ranging from 0 to 3 (0: none, 1: punctuate, 2: early confluent and 3: confluent).13 We used the term lacune to describe scars infarctions and referred to deep, subcortical or pontine ovoid lesions (3–15 mm), with cerebrospinal fluid-like signal with or without hyperintense FLAIR border.14 We scored lacunes as present or absent. We defined cerebral microbleeds (CMBs) as small round foci of hypointense signal in T2*-GRE-weighted images, of 10 mm or less in the brain parenchyma. We rated CMBs with the Brain Observer MicroBleed Scale.15 We scored CMBs as strictly lobar, strictly deep or mixed.16 We defined cortical superficial siderosis (cSS) as a homogeneous curvilinear signal intensity (black) on T2*-GRE sequences in the superficial layers of the cerebral cortex, within the subarachnoid space, away from at least two sulci of the haemorrhage with no corresponding signal hyperintensity on FLAIR sequences.17 We categorised cSS as focal (when restricted to three sulci or less), or disseminated.17 We classified cerebral atrophy using a simplified version of a validated scale,18 ranging from 0 to 3 (0: absent, 1: mild, 2: moderate, 3: severe).
Follow-up
As part of our standard in-house protocol for post-stroke care, patients were invited to be followed up at 6 months and 1, 2, 3, 4.5, 6, 8 and 10 years after the index ICH, as previously detailed.7 At each visit, the mRS was evaluated by a vascular neurologist. In order to describe the temporal trajectories of mRS over time when an mRS value was not available at an intermediate time point, we used the worse mRS score among the two closest visits, according to the ‘worst mRS scenario’ method.19 During the follow-up visits, we recorded the occurrence of any new stroke (ischaemic or haemorrhagic) and dementia, as previously described.7 11
Statistical analysis
Quantitative variables were expressed as means (SD) in the case of normal distribution or medians (IQR) otherwise. Categorical variables are expressed as numbers (percentage). Normality of distributions was assessed using histograms and the Shapiro-Wilk test. To evaluate differences in terms of baseline characteristics between patients with and without MRI, we calculated absolute standardised differences (ASD), ASD higher than 20 means a significant difference between groups. We first examined the clinical and MRI characteristics associated with functional decline, defined as a transition from mRS grades 0–3 to 4–5, by using univariate cause-specific Cox proportional hazard models for interval-censored data, considering death without transition as competing event. Cause-specific HRs (CSHRs) for functional decline were derived as effect sizes with their 95% CIs. We assessed the log-linearity assumption for quantitative predictors by examining Martingale residual plots and by using restricted cubic spline functions, and the proportional hazard assumption for each potential prognostic factor by examining the Schoenfeld residuals. Clinical predictors associated with functional decline (at a p value of <0.20 in univariate analyses) were implemented into a backward stepwise multivariable model using a value of p<0.05 as cut-off for retention in the final model. A second backward stepwise multivariable model was implemented by including MRI predictors associated with functional decline (at p value of <0.20 in univariate analyses) and independent clinical predictors identified in previous multivariable model. Finally, we analysed the impact of new clinical events (dementia and any stroke (combined new ICH and ischaemic stroke)) on functional decline by using semiparametric proportional hazard models for interval-censored data (by considering death without transition as competing event) with time-dependent covariates.20 Time-dependent covariates associated with functional decline (at p value of <0.20 in univariate analyses) were included in third backward stepwise multivariable models. Before developing multivariable models, the collinearity between the candidate predictors was appreciated by calculating the variance inflation factors with an alert threshold value of 2.5.21
All statistical tests were done at the two-tailed α level of 0.05. Data were analysed using SAS V.9.4.
Results
Characteristics of the study population
Of the 560 patients with spontaneous ICH enrolled in the PITCH cohort, we included the 174 (31%) patients alive at 6 months with mRS scores 0–3. Their baseline characteristics are reported in table 1: 66 patients were women (37.9%). Their mean age was 64.6 years (SD: 13.5) and their median National Institutes of Health Stroke Scale score at admission was 6 (IQR: 3–11). Three included patients were demented at 6 months from index ICH.
Cause-specific hazard models: univariate associations of clinical and ICH characteristics, and new clinical events with functional decline
mRS changes over time
The breakdown of mRS in these 174 patients is reported in table 1 and figure 1. During a median follow-up of 9 years (IQR: 8.1–9.5), 40 patients (23%) converted to mRS scores of 4–5. For each mRS baseline score, trajectories during follow-up are depicted in figure 2.
Modified Rankin Scale (mRS) changes over time. On the Y axis: time of each follow-up visit ranging from 6 months (M6) to 10 years (Y10) after the index intracerebral haemorrhage (ICH). On the X axis: percentages of patients. Legend colour for each mRS category is specified in the figure. On the right is the reported number of patients at each time point.
Modified Rankin Scale (mRS) trajectories for each baseline score. mRS trajectories are described with a Sankey diagram. Each panel depicts the mRS score trajectories during follow-up for each mRS category evaluated at the 6 months’ visit (M6). The width of the flow line is proportional to the number of patients. Darker colours depict trajectories of patients with functional decline or death during follow-up. Numbers in the flow line represent the mRS category. On the bottom of each panel is represented the time of each follow-up visit from M6 to 10 years (Y10) after the index intracerebral haemorrhage.
Clinical and ICH characteristics associated with functional decline
Univariate associations of baseline characteristics with transition from mRS 0–3 to 4–5 are presented in table 1. In multivariable analysis including all the variables associated in univariate analyses with transition from mRS grades 0–3 to 4–5 (with p value of <0.20), we found that age (CSHR per 10-year increase: 1.08; 95% CI: 1.05 to 1.12), diabetes mellitus (CSHR: 2.49; 95% CI:1.18 to 5.28), ICH volume (CSHR per 1 mL increase: 1.03; 95% CI: 1.01 to 1.06) and mRS score at 6 months (CSHR per 1 point increase 1.72, 95% CI 1.24 to 2.41) were independently associated with transition from mRS 0–3 to mRS 4–5 (table 2, model 1).
Cause-specific hazard models: multivariable associations of baseline clinical and MRI characteristics with functional decline
Small vessel disease MRI markers associated with functional decline
Thirty (17%) patients had contraindications for MRI (mainly pacemakers, claustrophobia and unstable clinical state). Differences in terms of baseline characteristics between patients who were able to undergo MRI and those who were not are reported in online supplemental table 1.
Supplemental material
Univariate associations between MRI markers and functional decline are presented in table 3.
Cause-specific hazard models: univariate associations between MRI characteristics and functional decline
In multivariable analysis, including all MRI markers associated with transition from mRS 0–3 to 4–5 in univariate analyses and independent clinical factors selected in the previous clinical multivariable model, we found that presence of strictly lobar CMBs (CSHR: 3.46; 95% CI: 1.15 to 10.45 compared with no CMB), and mixed CMBs (CSHR: 4.33; 95% CI: 1.68 to 11.16 compared with no CMB) were independently associated with functional decline in addition to age (CSHR: 1.07, 95% CI: 1.03 to 1.11), diabetes mellitus (CSHR: 2.54; 95% CI: 1.07 to 6.02), ICH volume (CSHR per 1 mL increase: 1.04; 95% CI: 1.01 to 1.07) and mRS score at 6 months (CSHR per 1 point increase 1.60, 95% CI 1.10 to 2.32; table 2, model 2).
Occurrence of new clinical events during follow-up
Outcome of interest (mRS score) during follow-up was available for all patients. During the follow-up, 48 (28%) patients developed dementia, 16 (9%) ischaemic strokes and 8 (5%) ICH. New clinical events during follow-up were all significantly associated with an increased risk of transition to mRS 4–5 (table 1).
When occurrence of dementia and any stroke was added to model 2 including baseline clinical and MRI characteristics independently associated with functional decline, age (CSHR: 1.07; 95% CI: 1.03 to 1.11), ICH volume (CSHR: 1.03; 95% CI: 1.01 to 1.06), mRS score at 6 months (CSHR per 1 point increase 1.61, 95% CI 1.07 to 2.43), occurrence of dementia (CSHR: 3.81, 95% CI: 1.78 to 8.16) and occurrence of any stroke during follow-up (CSHR: 4.29, 95% CI: 1.80 to 10.22) remained independently associated with transition to mRS 4–5.
Discussion
In a prospective cohort of 174 consecutive ICH survivors with a relatively favourable outcome (mRS 0–3) 6 months after ICH, we showed that one patient out of four significantly declined (conversion to unfavourable outcome (mRS 4–5)) in approximately 9 years of follow-up. Clinical characteristics associated with functional decline were older age, diabetes mellitus, larger ICH volumes and higher mRS score at 6 months. Among MRI markers of chronic cerebral injury evaluated at baseline, presence of strictly lobar CMBs and mixed CMBs were also independently associated with functional decline. However, when occurrence of new clinical events during follow-up were entered in multivariable models, only age, baseline ICH volume, mRS score at 6 months, dementia and new strokes after ICH remained independently associated with functional decline.
The major strengths of our study are (1) the duration of the follow-up that is the longest to date in prospective hospital-based ICH studies; (2) a detailed, standardised and complete assessment of MRI characteristics of cerebral small vessel disease (SVD) according to current recommendations14; (3) prospective assessment of new significant clinical events such as dementia, ischaemic stroke and ICH through face-to-face visits. This study has also limitations. First, this is a single-centre study limiting the generalisability of our results. However, the baseline characteristics of the PITCH cohort are similar to those of a population-based ICH study of the same country suggesting that recruitment bias was minimised.10 We acknowledge that our sample size might be relatively limited. Furthermore, we acknowledge a relative small number of outcome events that could have influenced the results of our multivariable analyses. However, for the purpose of our study, we have restricted our analyses to patients with at most a minor disability at 6 months of the index ICH. Clinical data on this category of patients with ICH are lacking and are valuable for both clinicians and researchers as they represent the preferable target population for investigational therapeutic interventions in the chronic phase of ICH. As commonly seen in studies evaluating ICH survivors, we acknowledge that, in a minority of patients, MRI scans were not available. To inform potential selection bias, we report differences in terms of clinical characteristics in supplemental materials. In ICH cohort, death occurring during follow-up can be a serious competing risk to evaluate outcome measures such as functional decline. Therefore, we chose to use competing risk models to address this concern. We also acknowledge as a limitation the absence of a control group.
Our study provides novel data relevant for both clinical and research setting showing that almost one-fourth of patients with minor disability 6 months after ICH will convert to major disability during follow-up. Furthermore, we evaluated mRS trajectories during 10 years of follow-up. As mRS is used as the principal endpoint in most clinical trials evaluating therapeutic interventions in patients with ICH, temporal trajectories of functional decline over time provide useful information for the design of future trials.6
We found that several baseline demographics and clinical characteristics are associated with functional decline such as age, diabetes mellitus and ICH volume. Furthermore, we found that higher mRS scores at 6 months (from 0 to 3) were strongly associated with functional decline during follow-up. This might also be visually appreciated by mRS trajectories depicted in figure 2. As baseline MRI markers of SVD, we found that patients with both strictly lobar and mixed CMBs were more likely to suffer from significant functional decline during follow-up. The relationship we found between strictly lobar CMBs, mixed CMBs and functional decline suggests that CAA might be a potential contributor to functional decline. However, some studies found that the concomitant presence of lobar and deep haemorrhages (ie, mixed CMB) was associated to a diffuse and severe form of SVD.9 Therefore, our results might suggest that also a more severe form of SVD-related parenchymal damage might be an important determinant of functional decline.
We evaluated the association between occurrence of dementia and functional decline during follow-up. The criteria used for diagnosis of dementia required cognitive or behavioural symptoms that interfere with the ability to function at work or at usual activities with a potential partial overlap with functional decline.22 This choice was made as we aimed to specifically assess the influence of cognitive impairment on functional decline after ICH. In line with our analyses, evidences from studies that evaluated patients with mild to moderate Alzheimer’s disease suggest that cognitive impairment precedes and predicts functional impairment.23 We found that clinical events after ICH such as dementia and stroke are major determinants of functional decline. In fact, when clinical events during follow-up were added into the multivariable models, part of the previously mentioned baseline clinical and MRI markers were no more significant. One possible interpretation is that MRI markers of CAA and SVD burden (ie, strictly lobar CMBs and mixed CMBs) were associated with a major risk of developing major disabilities during follow-up through their effect on increasing dementia risk and ICH recurrences.7 9 24 This might also be true for diabetes mellitus that has been associated to cognitive impairment.25 Our results have also important clinical implications as they confirm the paramount importance of finding new interventions aiming at lowering dementia and stroke risks after ICH. Furthermore, they highlight the importance of the ongoing clinical trials that evaluate the best therapeutic option in prevention of thromboembolic events after ICH.26
In conclusion, in 6-month ICH survivors with a relatively favourable outcome, one patient out of four will develop functional decline during follow-up. MRI markers of CAA and severe SVD burden evaluated at baseline are associated with higher risks of functional decline. However, when new clinical events during follow-up were entered in the model, only age, ICH volume, mRS score at 6 months, new dementia and stroke remained associated with functional decline after ICH.
Acknowledgments
CC is a member of the Institut Universitaire de France.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @marco_pasi85, @BarbaraCasolla, @gboulouis, @leysdidier1
Contributors MP designed and conceptualised the study, analysed and interpreted all data and drafted the manuscript. BC and SM interpreted all data and revised the manuscript. MK and JL performed the statistical analyses. GB and GK contributed to data collection, analysed imaging data and revised the manuscript. HH conceptualised and designed the study, interpreted study data and revised the manuscript. DL and CC designed and conceptualised the study, analysed and interpreted all data, and drafted the manuscript.
Funding This study was funded by Inserm U1172 and Adrinord.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.