Functional recoveries of patients with branch atheromatous disease after rehabilitation: Comparison with other types of cerebral infarction and importance of stratification by clinical categories

1.1 Prediction of functional outcomes in patients with stroke

Accurate and early prediction of outcomes after stroke is important for optimizing rehabilitation management. There are many reports concerning the prognosis of patients with stroke after rehabilitation. They have demonstrated that outcomes and/or functional recoveries after stroke are influenced by motor and cognitive functions on admission, length of hospital stay, age (Branco et al., 2019; Jørgensen et al., 1995b; Kelly et al., 2003; Kimura et al., 2019; Kwakkel & Kollen, 2013; Lin et al., 2019; Meyer et al., 2015), gender (Paolucci et al., 2006), amount and/or intensity of rehabilitation (Kamo et al., 2019; Miyai et al., 2011; Okuda & Nakata, 2020; Umehara et al., 2018), timing of the initiation of rehabilitation (Imura et al., 2018; Yagi et al., 2017) and quality of rehabilitation (Kinoshita et al., 2015; Langhammer et al., 2017; Sonoda et al., 2004).

From daily clinical observation, we assumed that the type of stroke might be one of essential factors to determine the functional recovery. Although differences in neurological outcomes and functional outcomes elicited by the cause of stroke (intracerebral hemorrhage vs cerebral infarction) remain controversial (Jørgensen et al., 1995a; Paolucci et al., 2003), the functional recovery may be dependent on stroke types (Kelly et al., 2003; Kitago & Ratan, 2017; Stabel et al., 2017). Indeed, a few reports suggested that the degree of recovery after cerebral infarction varied according to the pathogenic mechanism (Niimi et al., 2016; Paci et al., 2011).

1.2 Stratification of patients with cerebral infarction by clinical characteristics

Clinical categories proposed by the NINDS (Classification of Cerebrovascular Disease III) have commonly been used for classification of stroke (Special report from the National Institute of Neurological Disorders and Stroke, 1990). We followed this classification and categorized cerebral infarction into atherothrombotic, cardioembolic, lacunar and others, based on the pathogenic mechanism. Intracranial branch atheromatous disease (BAD) was first described in 1989 and has been recognized as a type of ischemic stroke (Caplan, 1989). It is clinically characterized by frequent early neurological deterioration, and often occurs in the area perfused by a penetrating artery from the middle cerebral artery (MCA) or a paramedian penetrating artery of the basilar artery (BA). It was originally defined as having a long axis of more than 10 mm in the cerebral hemisphere, and as reaching the ventral surface in the brain stem. Neuroradiologically, BAD has frequently been defined to have a lesion with a long axis or depth of more than 15 mm in the cerebral hemisphere, or a paramedian pontine lesion reaching the ventral surface of the brain stem with diffusion weighted MRI, although a review article in 2016 found little consistency about the definition of BAD (Petrone et al., 2016).

1.3 Functional outcomes of BAD after rehabilitation

In spite of the high incidence of early neurological deterioration, functional recoveries after rehabilitation of patients with BAD have not been well documented, and there have been only a few studies available concerning long-term outcomes of patients with BAD (Niimi et al., 2016; Senda et al., 2013; Suto et al., 2009). From our daily observation of patients, we hypothesized that patients with BAD might have better functional improvement among patients with different types of cerebral infarction, and further thought that we might have advantages in comparing the functional recoveries of patients with different types of cerebral infarction, because we have retained consistent diagnostic processes, rehabilitation programs and criteria for admission and discharge in the past 10 years in our institution. We, therefore, analyzed outcomes of patients with BAD after rehabilitation, and compared with functional recoveries of patients having other types of cerebral infarction, namely cardioembolic cerebral infarction (CE) and atherothrombotic cerebral infarction (AT).

2.1 Design of the study

This retrospective observational study included consecutive patients with cerebral infarction, who were discharged from Tane Neuro-Rehabilitation Hospital, located in the western region of metropolitan Osaka, Japan. This study has been approved by the Institutional Review Board of Tane Neuro-Rehabilitation Hospital (Approval Number 2019112; Approval Date September 28, 2019), and comprehensive consent was obtained on admission, from each patient and/or family members in a written form.

2.2 Measures used for this study

Functional independence measure (FIM), which evaluated activity of daily living (ADL), was used as a measure of function. FIM on discharge was evaluated as an index of functional outcome, whereas FIM effectiveness was evaluated as an index of functional improvement by using the formula; (FIM on discharge–FIM on admission)/(126–FIM on admission).

We chose FIM effectiveness over absolute FIM gain (FIM on discharge –FIM on admission) for evaluating functional improvement, because FIM gain has a ceiling effect and does not take the potential maximal functional improvement into account. After the proposal to use rehabilitation effectiveness as an index of rehabilitation outcome (Heinemann et al., 1987; Shah et al., 1990), FIM effectiveness has been used for evaluation of the functional improvement after rehabilitation (Gialanella et al., 2011; Koh et al., 2013; Ng et al., 2016; Spaccavento et al., 2017; Tokunaga et al., 2017; Tsujimoto et al., 2020).

2.3 Patients

Patients with cerebral infarction who were discharged from our hospital between January 2010 and December 2018 (N = 688) were divided into 5 groups with reference to the NINDS category, CE (N = 177), AT (N = 219), lacunar infarction (N = 28), others (N = 18), and BAD (N = 222). Patients who were transferred to other institutions because of interposing serious illnesses or neurological comorbidities were excluded (N = 24), and patients who were classified to have lacunar infarction and others were also excluded from further analyses, because of the limitation in numbers.

The diagnosis of BAD was based on the following neuroimage findings in the acute phase; a paramedian pontine lesion by the diffusion weighted MRI together with a corresponding low apparent diffusion coefficient, which reached the ventral surface of the pons in the territory of paramedian perforators of the BA, or a lesion encompassing the basal ganglia, internal capsule, and/or corona radiata by the diffusion weighted MRI together with a corresponding low apparent diffusion coefficient, which was more than 15 mm in the long axis or depth in the territory of perforators of the MCA. Patients were admitted to our hospital, when they completed acute medical care and achieved a stable general condition. Discharge was decided when patients were considered to have reached a plateau in ADL, regardless of their discharge destination.

2.4 Analyses

All 618 patients with BAD, CE, and AT were consecutively subjected to comparison of gender, age, location of the brain lesion (supratentorial vs infratentorial), side of the brain lesion (right vs left), time to admission after onset, motor FIM on admission, cognition FIM on admission, therapy time per day, length of stay in hospital, FIM on discharge as an index of outcome, FIM effectiveness as an index of functional improvement and the ratio of patients discharged to home. These factors were analyzed using non-repeated measure analysis of variance (ANOVA) with Bonferroni correction or Kruskal Wallis H-test with Sheffe to avoid type I errors caused by multiple comparisons.

To investigate the effect of each factor as described below on functional improvement, FIM effectiveness was subjected to stratified analysis of gender (male vs female), age (< 65, 65–72, 73–80, > 80 years), location of the brain lesion (supratentorial vs infratentorial), side of the brain lesion (right vs left), time to admission after onset (< 22, 22–27, 28–36, > 36 days), motor FIM on admission (33, 33–53, 54–74, > 74) and cognition FIM on admission (< 15, 15–24, 25–32, > 32). Median and interquartile ranges were calculated for each subdivided component (subgroup) of each factor, encompassing all patients across three types of cerebral infarction (N = 618). FIM effectiveness was compared among three types of cerebral infarction by non-repeated measure ANOVA with Bonferroni correction.

As the final step of analyses, stepwise multiple regression analysis was carried out to evaluate the impact of suspected predicting factors for outcomes and improvement of ADL for each type of cerebral infarction separately. In these analyses, the dependent variable was FIM on discharge or FIM effectiveness, whereas independent variables included gender, age, location of the brain lesion, side of the brain lesion, time to admission after onset, motor FIM on admission, cognition FIM on admission, therapy time per day and length of stay in hospital. Binary data were incorporated as dummy variables for gender (male: 0, female: 1), side of the lesion (right: 0, left: 1), and location of the lesion (supratentorial: 0, infratentorial: 1).

Statistical analyses were performed using Bell Curve for Excel. P–values < 0.05 were considered to be statistically significant.

3.1 Univariate analysis

Characteristics of patients with each type of cerebral infarction are presented in Table 1. Time to admission after onset was significantly shorter, and cognition FIM on admission was significantly higher in patients with BAD compared to patients with CE or AT. On the other hand, no significant difference was observed among them in the ratio of gender, laterality of the lesion, motor FIM on admission or therapy time per day. Patients with AT were younger on admission than those with BAD or CE. The ratio of the supratentorial to infratentorial lesion was significantly higher in patients with CE than those with BAD or AT. As for outcome indices, no difference was observed among three types of cerebral infarction in length of stay in hospital, FIM on discharge or the ratio of discharge to home, even though FIM effectiveness in patients with BAD was significantly higher than those with CE or AT.

Stratified analysis of FIM effectiveness for patients with each type of cerebral infarction is summarized in Table 2. There was no significant difference in FIM effectiveness among subgroups of patients categorized by side of the brain lesion, time to admission after onset, motor FIM on admission, or cognition FIM on admission. Patients with younger age, shorter time to admission after onset, higher motor FIM on admission or higher cognition FIM on admission had higher FIM effectiveness in all three types. On the other hand, FIM effectiveness was significantly higher in male patients with BAD compared to male patients with CE or AT. The difference in FIM effectiveness between male and female patients was clear in BAD, but it was not seen in CE or AT. FIM effectiveness was also significantly higher in patients with BAD, who had supratentorial lesions compared to those with supratentorial CE or AT lesions. In the subgroup under 72 years old, FIM effectiveness in patients with BAD showed a trend to be higher than those under 72 years old with CE or AT.

Stepwise multiple regression analysis for FIM on discharge (Table 3) showed that it was significantly influenced by motor FIM on admission, cognition FIM on admission and length of stay in hospital in all three types of cerebral infarction. This analysis also showed that location of the brain lesion (supratentorial vs infratentorial) and gender (male vs female) significantly influenced FIM on discharge in patients with BAD, but not in those with CE or AT. Stepwise multiple regression analysis for FIM effectiveness (Table 4) showed that it was significantly affected by motor FIM on admission and age in all three types of cerebral infarction. However, this analysis also showed that gender (male vs female) was a significant predicting factor in patients with BAD, but not in those with CE or AT.

4.1 Differences in functional recoveries among three types of cerebral infarction

This retrospective observational study demonstrated differences as well as similarities of outcomes after rehabilitation of patients with BAD, CE and AT by using univariate, stratified and multivariate analyses. Patients with BAD had functional outcomes comparable to CE and AT after rehabilitation, as estimated by FIM on discharge, and better functional improvement, as evaluated by FIM effectiveness, compared to those with CE or AT. The factors related to functional outcomes were similar to each other, but those related to functional improvement were different among BAD, CE and AT. Although the predicting factors for FIM effectiveness in stroke patients are still unsettled in previous reports (Ng et al., 2016; Spaccavento et al., 2017; Tsujimoto et al., 2020), the present study suggested that the functional improvement after rehabilitation of patients with cerebral infarction should be evaluated according to their clinical types.

Consistent with previous studies, a better functional status including high cognition FIM and motor FIM on admission as well as longer stay in hospital were strongly associated with better functional outcomes after rehabilitation of patients with cerebral infarction regardless of clinical types. A lack of difference in functional outcomes among patients with BAD, CE and AT, as evaluated by FIM on discharge, may be explained by the similarity of their characteristics on admission including motor FIM on admission and length of stay in hospital (Table 1) and by the similarity of factors related to functional outcomes (Table 3).

4.2 Characteristics of functional recoveries of patients with BAD

In contrast, better functional improvement, as shown with higher FIM effectiveness, was present only in patients with BAD compared to those with CE or AT (Table 1). Considering the previous studies showing association of higher cognitive function and earlier initiation of rehabilitation with better recovery of ADL after stroke (Branco et al., 2019; Imura et al., 2018; Kimura et al., 2019; Meyer et al., 2015; Tsujimoto et al., 2020; Yagi et al., 2017), a shorter time to admission after onset of stroke and higher cognition FIM on admission (Table 1) shown in our patients with BAD might have contributed to better functional improvement.

Stratified analysis demonstrated significantly better functional improvement in male patients with BAD compared to those with CE or AT (Table 2). Multiple regression analysis also showed gender to be a predicting factor for functional improvement in patients with BAD, but not in those with CE or AT (Table 4). Thus, gender should be an important predicting factor, when one estimates the degree of functional recovery, particularly in patients with BAD.

Location of the brain lesion was also identified as a predicting factor for functional outcomes of patients with BAD (Table 3), where functional improvement of patients with supratentoeial BAD was better than patients with supratentorial CE or AT lesions (Table 2). Functional recoveries of stroke patients may be influenced by location (supratentorial vs infratentorial) or side (right vs left) of brain lesions (Gialanella et al., 2011; Kawano et al., 2018; Spaccavento et al., 2017; Tsujimoto et al., 2020). As anticipated from neuroanatomy, patients with BAD who have infratentorial lesions tend to have ataxia (Kumral et al., 2002), which may lead to lesser functional improvement compared to those who have supratentorial lesions.

As expected from previous studies, younger age, shorter interval from onset to admission, higher motor function and higher cognitive function on admission lead to better functional improvement more or less in all three types of cerebral infarction (Tables 2, 4). However, the impact and the type of predicting factors related to FIM effectiveness in patients with BAD were different from those in patients with CE or AT. Age and cognitive function on admission may exert more influence on functional improvement of patients with BAD, because age-dependent difference was clearly detected in FIM effectiveness (Table 2) and standardized coefficient of cognition FIM on admission was highest along with motor FIM on admission among variables for FIM effectiveness of patients with BAD (Table 4).

Taken together, outcomes after rehabilitation of patients with BAD may be characterized by better functional improvement if patients are male, are relatively younger, or have supratentorial lesions in comparison to patients with other types of cerebral infarction. Better improvement in patients with BAD may be influenced by better cognitive function on admission compared to patients with CE or AT.

4.3 Limitations of the study

We investigated functional recoveries of patients with cerebral infarction by using FIM on discharge as an index of functional outcome and FIM effectiveness as an index of functional improvement. In multivariate analysis, adjusted R-squared of FIM effectiveness was low (0.28–0.40), whereas that of FIM on discharge (0.81–0.85) was high. Those results suggested that the clinical factors used in this study could be useful for predicting functional outcomes of cerebral infarction, but not functional improvement well. More candidate factors based on comorbidities, life style or genetic predisposition are assumed to be the individualized predictive approach for stroke (Polivka, et al., 2019). New biomarkers such as cell-free nucleic acid and miRNA have been shown to be associated with diagnosis, severity and outcome of ischemic stroke (Crigna et al., 2020), and better factors may contribute to precise prediction of functional recoveries after stroke in the future.

In addition, our findings may not be applicable generally to every patient with ischemic stroke, because there may be potential biases caused by a relatively small sample size, caused by the data from a single medical center, or caused by underlying confounding factors in three types of cerebral infarction. Conversely, the data were relatively robust without variation of some factors such as diagnostic processes, treatment plans, and rehabilitation programs. Despite the limitations mentioned above, we thus believe the results are meaningful.