Minor stroke is receiving an upsurge of attention in China. The CHANCE study showed that a combination of clopidogrel with aspirin is superior to aspirin alone for protecting against subsequent stroke in new TIA and minor stroke patients [10]. However, this study did not categorize the patients according to etiology, which is critical to individualized therapy. As combining the results of randomized controlled trials with individualized considerations may have therapeutic benefits, we believe that it would be beneficial for minor stroke patients to identify their underlying etiology.
The TOAST system is the most standard etiological classification system for ischemic stroke. As medical technology and clinical studies progress, the pathogenesis of ischemic stroke becomes more concrete. The CISS system represents a new etiological system for ischemic stroke [8]. A significant decline in subtypes of LAA and SVD in stroke/transient ischemic attack has been found in Western countries like England and Canada [11]. However, LAA, especially intracranial atherosclerosis (ICAS), is more common in Asian societies [12]. Furthermore, a high proportion of ICAS patients who experienced a minor stroke were at high risk of developing early neurological deterioration (END) [13]. The CISS system, which further classifies the underlying mechanism of LAA, might therefore be more appropriate for use with Chinese patients with minor stroke. However, the reliability of CISS has not been assessed. Here, we compared the CISS and TOAST systems in patients of minor stroke and examined reliability scores for them.
CISS-etiologic category of minor stroke patients
In our study, 320 minor stroke patients were divided into five primary categories, consistent with the diverse causes and mechanisms known to underlie strokes. According to CISS, LAA was the largest subtype, which may be explained by the high prevalence of ICAS in China [14, 15]. However, CS was less than 5 % in our study, which is partly because most cardiac embolisms cause severe neurological deficits [16, 17] that results in an NIHSS score of above 3 on admission. Furthermore, there may have been limitations in our diagnostic tools. For example, transesophageal echo (TEE) and 30 events monitor or loop monitor are not routine procedures for ischemic stroke patients in our hospital.
Distribution of etiology-CISS compared with TOAST
In our study, minor stroke patients were classified into diverse groups according to the etiology and mechanism of the disease. We found mismatches in LAA and UE with CISS and TOAST. In the TOAST system, most patients were classified into SAO, followed by UE and LAA, whereas in CISS, LAA and PAD groups took the largest proportion. This may reflect some differences in the definitions of subtypes across the two systems.
In our study, about half of the patients that were classified into SUE in TOAST were ascribed to LAA in CISS (52/97 in Neurologist A’s results, 38/82 in Neurologist B’s results). TOAST explicitly defines LAA with both specific stenosis degree of parent artery and size of lesion. As a result, if no other cause was found, the following two conditions would be classified as SUE using the TOAST classification: stenosis degree ≤ 50 % with the lesion diameter ≥ 1.5 cm; stenosis degree >50 % with the lesion diameter < 1.5 cm. By contrast, CISS does not include a restriction for stenosis degree or lesion diameter. Additionally, more than a quarter of the patients classified into SUE in TOAST were ascribed to PAD in CISS (27/97 in Neurologist A’s results, 25/82 in Neurologist B’s results). To help differentiate from lacunar stroke, CISS proposed the notion of PAD, as caused by atherosclerosis at the proximal segment of the penetrating artery or lipohyalinotic degeneration of an arteriole. The diagnosis of PAD was established for isolated penetrating artery territory infarct, when there was no evidence of atherosclerotic plaque or any degree of stenosis in the parent artery. The diagnostic procedure ignores the lesion size and clinical manifestation. Thus, compared with TOAST, CISS significantly decreased the number of patients classified as SUE, and might be more useful for clinicians to assess the etiologies and mechanisms of patients with minor stroke.
Differences between SAO and LAA were found to be the second discrepancy between TOAST and CISS, likely attributable to the different definitions of LAA used by the two systems. In TOAST, isolated PA territory infarct is classified into LAA only when both of the following conditions are included: i) brain imaging findings of either significant (>50 %) stenosis or occlusion of parent artery; and ii) the lesion is not smaller than 1.5 cm in diameter. By contrast, CISS further classifies LAA into four categories according to the underlying mechanism and emphasizes the significance of atherosclerotic plaque other than the lesion size. This means that proof of atherosclerotic plaque, any degree of stenosis in the parent artery, or new isolated PA territory infarcts should all be ascribed to LAA. Indeed, when other possible causes were excluded, patients with multiple small lesions (diameter <1.5 cm) in cortical or subcortical regions were attributed to SAO with the TOAST criteria, but classified into LAA in CISS if all the lesions were in the territory of the stenosed artery. In fact, the potential mechanism of this kind of infarct is artery-to-artery embolism [18], a branch of LAA with CISS. Therefore, a portion of SAO patients in TOAST were classified as LAA in CISS.
Inter-rater reliability: CISS compared with TOAST
Both TOAST and CISS were found to be reliable in our study (inter-rater agreement: k = 0.732 for TOAST, k = 0.898 for CISS). CISS, which categorized the etiologic subtypes of minor stroke with excellent inter-examiner reliability based on assessment of clinical data obtained through medical record abstraction, showed higher practical utility. Although the TOAST classification has been widely used in prospective clinical trials and retrospective studies, its reliability was not always excellent or stable in previous studies [6, 19–21]. By contrast, the high inter-examiner agreement rates of the CISS suggest its potential utility in stroke research, though it should be noted that there are still some issues when applying CISS criteria. Differences in interpretation of medical records were a leading source of disagreement among examiners, and discrepant comprehension of subtype definition also contributed to the instability of CISS.
Implications for treatment: CISS Classification
CISS is a more detailed etiologic system for ischemic stroke than TOAST, which means that it might be more practical for individualized treatment. As mentioned before, the underlying mechanism of patients with low-grade stenosis and small lesions is artery-to-artery embolism and/or impaired emboli clearance with CISS. For them, the root is the instability of plaques or thrombi, so intensive statin therapy may be beneficial. With regard to hypoperfusion cases, hypervolemic treatment is essential. However, we have to acknowledge that whether the impact of CISS on the schemes of therapy and secondary prevention could change clinical outcomes is still uncertain. Large prospective studies may be the only way to reach a definite conclusion.
Limitations of study
There are some limitations in our research: (a) since we do not included the patients with moderate and severe strokes, our conclusions should not be generalized to the whole population of cerebral infarction; (b) as a retrospective study, bias is inevitable; (c) the insufficient diagnostic work-up could affect the accuracy of some results. For financial reasons, only a subset of patients took the examination of head computed tomography angiography (CTA), which is more precise in diagnosis of vascular stenosis. However, we do use high resolution MRA to increase the accuracy of the assessment. TCD emboli-monitoring and TEE are still not carried out in our hospital. The former is used to detect MES, especially in patients with a potential cardioembolic source, which is less common in minor stroke. Besides, a study with large cohort indicated that MES prevalence was relatively low in patients with a potential native cardioembolic source [22]. Therefore, the lack of TCD emboli-monitoring might not have a significant impact on our results. TEE could identify patent foramen ovale (PFO), although controversy still surrounds the issue of the relationship between PFO, paradoxical embolism, and cryptogenic stroke [23, 24]. Embolic Stroke is more likely to happen in PFO patients combined with venous thromboembolism [25, 26], which was not found in our objects. It therefore seemed reasonable to exclude the possibility that lack of TEE might significantly affect our results. Finally, we note there are some published studies that share the same insufficient diagnostic work-up with ours [20, 27]. In addition, a continuous heart monitor study could be required for definitive diagnosis of paroxysmal atrial fibrillation - a common cause of stroke, but for now it is difficult to implement because of the limited medical resource and economic cause.