Cryptogenic Stroke

Definition of Cryptogenic Ischemic Stroke

There is no formal consensus definition for cryptogenic stroke, and definitions have varied by the particular scheme used to classify ischemic stroke cause or mechanism (see section on classification schemes below). There is no gold standard for diagnostic criteria. There is, of course, no animal model of cryptogenic stroke pathophysiology, since it is fundamentally unknown. One widely used classification system, designed for the Trial of ORG-10172 for Acute Stroke Treatment (TOAST) trial, defined undetermined stroke as “brain infarction that is not attributable to a source of definite cardioembolism, large artery atherosclerosis, or small artery disease, despite extensive vascular, cardiac and serological evaluation.” As such, the definition is framed in negative terms, based on the absence of findings. An alternative proposed definition, based on infarct topography, is: “non-lacunar ischemic stroke of unknown etiology but suggestive of embolism despite an exhaustive search for etiologic cause,” the inference being that all non-lacunar ischemic strokes are due to embolism. Based on these considerations, a relatively recent proposal extends the use of infarct topography to categorize such patients as “embolic stroke of undetermined source” (ESUS). The main rationale for such an approach has been to define this group of patients in a positive manner, to enable a clearer definition for the conduct of randomized controlled trials, with, by extension, implications for clinical practice. Developing a consensus definition for cryptogenic ischemic stroke requires agreement on what is considered to be an extensive or adequate diagnostic evaluation and which findings are considered etiologic.

Classification Schemes for Ischemic Stroke Etiologic Subtypes

There are a number of classification schemes that seek to provide an objective and reliable framework for describing ischemic stroke etiology. In general terms, classification schemes include the domains of small-vessel disease, large-vessel atherosclerotic disease, cardiac causes of thromboembolism, unusual causes of stroke, and cryptogenic (also termed unknown, undetermined, or unexplained). The category of cryptogenic stroke often includes multiple groups, including those with unexplained stroke in the setting of extensive testing, stroke due to incomplete or inadequate diagnostic evaluation, and mixed or competing etiologies (e.g., concurrent carotid stenosis and atrial fibrillation). These subgroups within cryptogenic stroke are likely quite different in terms of the risks of future stroke and true underlying pathophysiology.

The three most prominent classification schemes are summarized below. While overall inter-rater agreement between these classification schemes is good to excellent, the percentage of patients classified as “undetermined” varies between schemes, ranging from 26% to 42% in one direct comparison. Each scheme utilizes information on clinical presentation, topographic features on neuroimaging, findings on imaging of large vessels, cardiac rhythm monitoring and imaging of the heart, and results of some serologic studies. However, there are differences in use of this information to assign subtypes:

• 1.

  Trial of ORG-10172 in Acute Stroke Treatment (TOAST) subtype classification system is the most widely used and has been incorporated into routine clinical practice, following initial development for clinical research purposes. The TOAST criteria categorize ischemic stroke into small-vessel occlusion, large-vessel atherosclerosis, cardioembolism, stroke of other determined etiology, and stroke of undetermined etiology. The definition of “undetermined” etiology includes three circumstances: the diagnostic evaluation is incomplete, no cause is found despite an extensive evaluation, or (most likely) cause cannot be determined because there is more than one plausible etiology found. Using the TOAST criteria, the inter-rater reliability of diagnosing stroke of undetermined etiology (i.e., cryptogenic stroke) is poor, which presents clinical and research challenges.

• 2.

  The Causative Classification Scheme (CCS) is a computerized algorithm that provides causative and phenotypic stroke subtypes, categorized as evident, probable, or possible supra-aortic large-vessel disease, cardioaortic embolism, small-vessel occlusion, other uncommon causes, or undetermined. Within the category of undetermined stroke, there is a further subtyping into unknown-cryptogenic embolism, unknown-other cryptogenic, unclassified, and incomplete evaluation. Of the three schemes discussed in this chapter, CCS appears to provide the lowest percentage classified as undetermined (26% vs. 39% for TOAST in the North Dublin Stroke Study, n = 381). Inter-rater reliability for this scheme is very good to excellent, because a low percentage of patients are labeled as unclassified, likely due to the use of a heavily probabilistic approach that more commonly assigns an alternative etiology than does other classification schemes.

• 3.

  Atherosclerosis, Small-vessel disease, Cardiac causes, Other, and Dissection (ASCOD): The ASCOD scheme categorizes patients with ischemic stroke into A: atherosclerosis, S:small-vessel disease, C: cardiac source, O: other cause, and D: dissection. The “other” category includes those disorders or findings that are uncommon, and is further subcategorized based on whether the finding is potentially causal, the causal link is uncertain, or the causal link is unlikely but an abnormality is present. The “other” category also includes patients for whom no cause is detected based on best available diagnostic tests and stroke-specific history. There is no specific category for undetermined or cryptogenic stroke, as these are included in the “other” category.

Probabilistic Approach to Etiologic Diagnosis

Common to all etiologic classification schemes, either stated or implied, is an assessment of probability that a finding is the putative cause of the stroke, through both direct and indirect inferences. First, identification of stroke etiology is usually based on identifying a probable source of thrombosis, rather than direct visualization of thrombus. For example, the presence of atrial fibrillation supports the diagnosis of cardioembolism, but thrombus in the left atrium or its appendage is only observed in a minority of patients. Second, the actual etiologic source may not be measured directly. For example, the diagnosis of small-vessel disease is not based on imaging small vessels but based on the assumption that small deep infarcts (<1.5 cm), especially in locations rich in perforator vessels, are probably due to occlusion of small vessels from processes that are inherent to the small vessel. Third, etiologies commonly coexist, requiring subjective judgment about which is most likely causal. Finally, some classification schemes assign a probability within etiologic subcategories, such as the CCS and ASCOD, because potential etiologic findings have differing levels of evidence to support a causal role.

Prevalence of Cryptogenic Stroke

The prevalence of cryptogenic ischemic stroke varies among studies, but estimates range from 15% to 35% of consecutive patients ( Table 44.1 ). Between-study comparisons are challenging and often unreliable, because of differences in etiologic classification scheme used, extent and quality of etiologic diagnostic testing completed, type of populations included (e.g., younger vs. older), and variations in attributing causality to some common findings (e.g., patent foramen ovale). Accordingly, valid inferences from between-study comparisons of temporal trends and regional variations in cryptogenic stroke are often not possible. Similarly, incidence cannot be reliably estimated from combining existing data. However, when using a standardized definition of ESUS, multinational studies have suggested that ESUS consistently accounts for about 16% of all ischemic strokes.

Factors Influencing Prevalence of Cryptogenic Stroke