Ischemic Lesion Location Based on the ASPECT Score for Risk Assessment of Neurogenic Dysphagia

Dysphagia is common among patients with acute ischemic stroke with a reported incidence of up to 78% [1]. It is associated with an increased rate of in-hospital complications (such as malnutrition and aspiration pneumonia) and with unfavorable long-term functional outcome [2, 3]. Therefore, early identification of patients with post-stroke dysphagia is of utmost importance in order to guide decisions concerning nutrition status, drug administration and necessity of further instrumental diagnostics (e.g., fiberoptic endoscopic evaluation of swallowing and/or videofluoroscopic swallowing study). Consequently, the current standard of stroke care requires that all patients should undergo a standardized dysphagia screening.

Although many different swallowing screening tests, which have been validated in the stroke population, provide acceptable sensitivity, no single test has been accepted as the gold standard yet [4‐9]. Hence, there is an urgent need to establish additional clinical parameters, e.g., radiological findings, to raise diagnostic accuracy in identifying dysphagic patients. Individuals with acute infarctions involving brain areas highly related to dysphagia could be targeted for expedited referral to a Speech and Language Pathologist (SLP) and/or instrumental swallowing assessment [10, 11].

Several studies have shown that volume, location, and lateralization of the lesion are associated with dysphagia incidence, severity, and characteristics of swallowing dysfunction [12‐17]. While right hemispheric stroke was linked to pharyngeal-stage dysfunction, aspiration, and higher dysphagia frequency, left hemispheric stroke was connected to oral-stage dysfunction and less severe dysphagia [13, 18, 19]. Furthermore, lesions in distinct brain areas such as the pre- and post-central gyri, opercular region, supra-marginal gyrus, and adjacent subcortical white matter tracts were identified to be related to dysphagia, with post-central lesions being especially associated with severe swallowing impairment [13]. Although considered to be a valuable information, the implementation of such radiological findings into clinical routine is challenging.

In this context, the Alberta stroke program early CT score (ASPECTS) can provide a tailored remedy: ASPECTS is a 10-point quantitative topographic CT scan score originally designed to quantify the degree of early ischemic changes in the brain parenchyma of patients with middle cerebral artery (MCA) stroke. By dividing the MCA territory in ten segments, it might allow for a standardized correlation between certain brain areas and dysphagia. A particular advantage of ASPECTS is its quick and easy applicability without the need for additional imaging processing steps, while offering simultaneously high reliability and reproducibility [20].

Given that evaluation according to the ASPECTS criteria is routinely performed in acute stroke care, a detailed investigation whether individually affected segments or the sum score might help to identify stroke patients at risk for dysphagia is of high practical importance. Since swallowing is a midline function with bilateral cortical activation, it remains controversial to what extent swallowing function is lateralized [21‐23]. Therefore, it would be particularly interesting to know whether differences regarding the association of affected ASPECT segments or the sum score exist between the right and the left hemisphere.

113 patients with an acute ischemic stroke in the MCA territory were included in this study. Mean age was 69 ± 13 years, and 40% were female. Baseline data and clinical variables of the study population are displayed in Table 1.

According to FEES, 62 patients (54.9%) were classified as having dysphagia. Dysphagic patients more often had aphasia, were mutistic, had an abnormal gag reflex, and suffered from buccofacial apraxia than non-dysphagic patients. NIHSS at hospital admission was significantly higher in the dysphagic group as compared to the non-dysphagic group (Table 1).

The frequency of dysphagia was not different between left- and right-sided MCA infarctions (40 out of 72 patients vs. 22 out of 41 patients; p = 0.500). Moreover, no difference in ASPECT sum scores was observed between strokes affecting the left or the right hemisphere (median 7 [IQR 5–9] vs. 7 [6‐9]; p = 0.579). In the left hemisphere the strongest associations between the ASPECTS segments and dysphagia were found for the lentiform nucleus (LN) (OR 0.113 [CI 0.028–0.433; p = 0.001), the insula (In) (0.275 [0.102–0.742]; p = 0.011), and the frontal operculum (M1) (0.280 [0.094–0.834]; p = 0.022; see Table 2 and heat map [Fig. 1]). For the right hemisphere, the strongest association was found for the insula region (0.385 [0.107–1.384]; p = 0.144), however, without statistical significance. In a stepwise multivariate regression model including all ASPECTS segments of the left hemisphere, an independent association with dysphagia persisted for the lentiform nucleus (OR 0.089 [CI 0.021–0.379; p = 0.001) and the insula (OR 0.209 [CI 0.066–0.660; p = 0.008), respectively. No independent associations were found for the right hemisphere.

Table 3 shows the relative frequency of dysphagia rising up to 100% with respect to the number and location (“hot spots”, see Table 2) of affected areas in patients with left hemispheric strokes as illustrated by the heat map.

ASPECT sum scores were lower in patients with dysphagia compared to patients without dysphagia (median 8 [IQR 6.0—9.0) vs. 7 [5.0–8.0]; p = 0.001; see Table 1). We found a steady increase of dysphagia risk with lower ASPECT scores for the left hemisphere. This was much less present in the right hemisphere (bar chart, Fig. 2).

Likewise, for unaffected “hot spots”, the risk of dysphagia increased with decreasing ASPECT scores; however, for affected “hot spots” a relevant risk of dysphagia already existed with higher ASPECT scores (see Suppl. Table 1a/b).

For the left hemisphere, multivariate logistic regression analysis including age and sex revealed lower ASPECT sum scores (OR 0.617 [CI 0.457–0.834]; p = 0.002) to be independently associated with dysphagia, whereas for the right hemisphere (0.738 [CI 0.528–1.032]; p = 0.076), this association was not present. Due to collinearity, the additional inclusion of the NIHSS score into the model abolished the predictive value of the ASPECT sum score in favor of the NIHSS (left hemisphere; OR for NIHSS 1.206 [CI 1.064–1.367]; p = 0.003).

The median Fiberoptic Endoscopic Dysphagia Severity Scale (FEDSS) in dysphagic patients was not different between the left and the right hemisphere (median 4.5 [IQR 4–5], n = 40 vs. 4.0 [2.75–5], n = 22; p = 0.578). We did not observe an association between ASPECTS segments and the severity of dysphagia.

A multivariate regression analysis including age and sex showed no significant association between lower ASPECTS score and pneumonia for both left hemispheric strokes (OR 0.802 [CI 0.626–1.028]; p = 0.081) and right hemispheric strokes (0.864 [CI 0.639–1.170]; p = 0.345). Moreover, a significant association between pneumonia and the three “hot spot” segments of the left hemisphere could not be observed.

In the right hemisphere, only the involvement of the insular cortex tended to be associated with swallowing dysfunction. In contrast, involvement of the lentiform nucleus, the insular cortex and the frontal operculum in the left hemisphere were predictive for dysphagia. Our findings are in line with results from previous studies. The inferior precentral gyrus and the insular cortex are known to represent critical nodes of the supratentorial deglutition network [15, 23, 33]. The insular cortex has been associated with related swallowing and nutritional properties such as coordination of oral musculature gustation, and autonomic functions, through connectivity or inherent ability of its own [34]. Its general involvement in swallowing has been consistently demonstrated over many studies, while findings are diverse as to whether there is a right or left hemisphere lateralization of this region [19, 35, 36]. Several authors suggest that the left insula strongly contributes to the initiation of the swallow by processing sensory elements and interacting with widespread brain regions in both hemispheres [23, 37, 38]. Although not as pronounced as for the insula, a left-sided lateralization of the frontal operculum during swallowing has been reported [33, 39, 40]. The strong association between the lentiform nucleus and dysphagia within the left hemisphere needs to be highlighted as an interesting finding. Although data on the role of the lentiform nucleus as part of the basal ganglia are scarce, neuroimaging studies have confirmed its involvement in swallowing [22, 35, 41]. To what extend lentiform nucleus lesions contribute to the development of post-stroke dysphagia remains unclear. Our data attribute a more profound role of the left lentiform nucleus in swallowing.

Looking at the frequency of dysphagia in those patients with afflicted “hot spot” areas (Table 3) in the left hemisphere, the involvement of the lentiform nucleus solely was associated with a relative dysphagia frequency of about 80%. Considering a combination of the lentiform nucleus with either the frontal operculum and/or the insula increases the relative frequency of dysphagia up to 100%. Those patients with infarction involving one or more of these ASPECTS segments (frontal operculum, insular cortex, lentiform nucleus) should be targeted for a more profound evaluation by a SLP or instrumental swallowing assessment independently from swallowing screening result.

We observed an inverse correlation between dysphagia risk and ASPECT sum score, which was much stronger for the left than for the right hemisphere (Fig. 2). As a potential explanation, a more widespread representation of swallowing within the left hemisphere can be assumed. This is in accordance with available literature describing greater anatomical connectivity for swallowing in the left hemisphere [23, 42].

We enriched the analysis by the additional evaluation of pneumonia incidence in our patients. A trend towards an association between lower ASPECT scores and pneumonia was observed for the left hemisphere.

The severity of neurological deficits in terms of NIHSS has been identified as an important clinical predictor of post-stroke dysphagia [3, 43, 44]. For example, a NIHSS score of > 13 is used as a triage instrument to facilitate the choice of an alternative feeding route within 48 h after stroke [45]. However, NIHSS scoring underlies a hemispheric bias as it over-represents comprehension-dependent (left hemisphere-associated) tasks rather than tasks relying on right hemisphere functions [46, 47]. For example, imaging studies have shown that patients with right hemispheric stroke can have low NIHSS scores despite substantial infarct volumes [46]. Here, ASPECTS enables a more objective assessment of stroke magnitude. Consequently, no difference was found for the ASPECT sum score between left- and right-sided MCA strokes.

Strengths of our study are the prospective recruitment of a homogeneous collective of patients with MCA infarctions. We consistently used FEES which is considered to be an objective swallowing assessment tool that offers an excellent diagnostic accuracy and inter-rater reliability in acute stroke patients to evaluate the presence of dysphagia in all patients in a time window of 24 h after hospital admission [25, 26, 48].

Shortcomings include the limited number of patients, thereby leaving the statistical evaluation of the right MCA infarction group underpowered. Moreover, the results for the frontal operculum (M1) have to be interpreted with caution, since the statistically significance did not withstand after adjustment for multiple comparisons. In order to accurately detect the extent of ischemic changes in CT scans according to ASPECTS, we analyzed the first CT scan with clear ischemic changes. Therefore, our findings might be limited regarding CT brain imaging in the very early phase of stroke, where ischemic changes still might be missing. Moreover, brain imaging consisted of both CT and MR scans, which might have caused heterogeneity in the determination of the ASPECT score. However, previous data showed that the differences between CT and MR imaging (DWI) in visualizing early infarction are small when using ASPECTS [32].

In summary, the distribution and extent of ischemic changes in the MCA territory according to ASPECTS may be used as risk indicator of neurogenic dysphagia in MCA infarction, particularly when the left hemisphere is affected. Although hitherto radiological findings are not frequently used in acute dysphagia management our data show that lesion location defined via ASPECTS provide additional predictive information. However, due to the exploratory nature of this research external validation studies of these findings are warranted in future. Nevertheless, we believe that this work might stimulate further research to transfer radiological parameters in acute dysphagia management.