Once an ischemic stroke has been identified, further diagnostic evaluation is needed to determine the cause, and thereby directs the management strategies employed to prevent another stroke. Several systems of sub-classification of ischemic stroke have evolved. The TOAST system leads to the single disease deemed to be directly and causally related to the ischemic stroke [2]. If none is found then the stroke is deemed cryptogenic. The ASCOD classification considers five potential etiologies of ischemic stroke: A for atherosclerosis, S for small vessel disease, C for cardiac source, O for other cause, and D for dissection [3]. When young to middle-aged patients with strokes who have been classified as cryptogenic by the TOAST criteria are then evaluated under the ASCOD system, multiple possible risk factors are commonly found. This fact emphasizes that it is possible to detect several underlying diseases or abnormalities that are not necessarily causally related to the index stroke thereby highlighting the degree of uncertainty that precludes a definitive assignment of causation in patients [4]. As the age of presentation with ischemic stroke increases, however, the majority of stroke patients demonstrate all three of the major potential categories of stroke causation: atherosclerosis, small vessel disease, and cardiac pathology, arguably strengthening the value of the ASCOD classification in guiding therapy in this patient subset [5].

A major mistake in the management of patients with an ischemic stroke is to immediately conclude that the presence of a PFO implies causation with a level of certainty that supports PFO closure. This false certainty may be mistakenly communicated to the patient as being curative by PFO closure. In addition, modification of other known stroke risk factors, risky behaviors, exercise, diet, and long-term antiplatelet treatment may be overlooked. Figure 11.1 provides an overview of different management strategies for patients with ischemic stroke and PFO.

The ASCOD classification system can be used to provide a framework for approaching patients with an ischemic stroke and a PFO. Referring to PFO as a disease or abnormality is controversial, as most in the cardiology community regard PFO as simply a remnant of the fetal circulation that is often innocent. However, the classification is useful in directing clinicians to consider the totality of findings in a given patient, the variability and difficulty of assigning levels of certainty to stroke causation, and the frequent presence of multiple potential etiologies.

Over the last few decades, clinicians and investigators have strived to understand the role of PFO in causation of ischemic stroke and the benefits versus risks of PFO closure. From these efforts, an alternative, more “PFO-centric” approach to the patient has evolved.

Although PFO is present in approximately 25 % of the general adult population, it has not been shown to increase the risk for stroke by itself [6]. Therefore, PFO closure should not be used as a primary stroke prevention strategy. In the case of secondary stroke prevention, it is critically important to evaluate patients with stroke and PFO with an eye to determining the level of suspicion that a paradoxical embolic event could reasonably account for the index stroke. In other words, is the PFO an incidental finding or a pathogenic one? Furthermore, even if paradoxical embolism is suspected in the presence of a PFO, major management questions still remain: Should the secondary prevention strategy be expanded to include transcatheter PFO closure? Should the medical treatment option be anticoagulation for a presumed venous-based thrombus rather than antiplatelet therapy?

These questions can be approached with an understanding of the fundamental logic that underlies the pathophysiology of PFO-related stroke and the use of PFO closure as a treatment strategy. The logic is based on the assumption that ischemic stroke is the result of a sequence of events:

The second key aspect of the logic is the distinction between coincidence (or association) versus causation. From the beginning, this distinction has been often overlooked. Coincidence is defined as a concurrence of events or circumstances without apparent causal connection. Causation can be defined as the production of an effect by a cause. This dichotomy is schematically represented in Fig. 11.2. Note that (1) the increased frequency of finding a PFO in cryptogenic stroke or other conditions does not prove causality and (2) the PFO is not a cause of stroke: PFO is a component of the causal nexus, i.e. the link between a cause and its effect. Using this clarity of organized thought and logical approach, the clinician then moves to the process of data gathering, interpretation, and decision-making. Departure from this logic associated with the pathophysiology of paradoxical embolism would occur if there was evidence of in situ thrombus in PFO’s with certain anatomical characteristics, that would make the PFO a primary cause of stroke. To date, however, such pathophysiology has remained speculative at best.

Stroke patients must undergo a comprehensive evaluation to gather data on relevant patient characteristics, clinical findings, laboratory values, index event characteristics, and cardiac and brain imaging variables. Figure 11.3 and Table 11.1 present an overview of the multiple pieces of data that are typically gathered in the evaluation of a patient with an ischemic stroke and a PFO. Categorizing these diverse data elements helps to organize the data, the interpretation of the findings, and the degree of certainty as to evidence that helps identify causation and direct therapy.

Currently there is not a comprehensive, validated model that integrates clinical, imaging, and other potentially relevant factors into a clinically useful tool to assist in decision-making. The prototype of such a clinical model is the development and validation of stroke risk scores (i.e., CHADS2 and CHADS2-VASc) used to assist decision making for anticoagulation in non-valvular atrial fibrillation [7, 8]. The score correlates with the risk of stroke and is a major factor used in the decision to treat patients with anticoagulants. A similar tool is needed in the PFO-stroke arena to assist in the selection of patients for PFO closure who have the highest risk for another stroke related to PFO-enabled paradoxical embolism and in whom medical therapy alone may not be as effective as when combined with PFO closure.

A clinical tool for PFO will have multiple key differences to tools used in atrial fibrillation. Atrial fibrillation is a common disorder with three central features: (1) a high likelihood of finding a causative thrombus in the left atrial appendage when a stroke has occurred; (2) if left untreated, a high likelihood of recurrent events especially when combined with other risk factors; and (3) the patient population is older and frequently has multiple risk factors and overt cardiovascular disease that heighten stroke risk both related to atrial fibrillation and other mechanisms. In contrast, with PFO related strokes there is: (1) a low chance of finding a causative thrombus; (2) a low risk of recurrent stroke; and (3) the patient population is usually younger and frequently has a few risk factors but no overt cardiovascular disease [9]. However, the quoted low risk of recurrent stroke in PFO patients was determined in patients on medical therapy, thus one can argue that the comparison to atrial fibrillation is not dramatically different. In addition, PFO closure may be considered in older patients but with the greatly increased issue of multiple competing stroke etiologies.

The Risk of Paradoxical Embolism (RoPE) model is a new development in the field of PFO and is an index score designed to risk stratify cryptogenic stroke patients with PFO [10, 11]. The likelihood that the stroke in a patient was related to their PFO is suggested by a high RoPE score, versus that the PFO was not part of the causal nexus, i.e. an incidental PFO, as suggested by a low RoPE score. This topic is further discussed in Chap.​ 20. The RoPE score method is predicated on the finding that the prevalence of PFO is higher (50–60 %) among CS patients than the general adult population (25 %). Using the difference between the baseline frequency of PFO and that of the likelihood of a PFO-related stroke for a given RoPE score, the higher the RoPE score, the greater the statistical “attributability” of the PFO to the cryptogenic stroke. The “attributable fraction” is estimated using Bayes theorem and a control rate of PFO found in 25 % of adults.

The RoPE database was derived from 12 published studies with a total of 3,674 cryptogenic stroke patients who had been investigated for PFO. Multivariable associations between predictors and the presence of a PFO were examined to build the model. The variables studied included patient clinical characteristic and radiologic variables from MRI or CT. The variables associated with a causative PFO in CS patients were the following: younger age, the presence of a cortical stroke on neuroimaging, and the absence of diabetes, hypertension, smoking, and prior stroke or TIA.

Two features of the RoPE score must be understood. First, causation is not proven with higher RoPE scores, but rather the association of the variables is greater with the presence of PFO. The central fallacy of cum hoc ergo propter hoc, correlation proves causation, must be kept in mind. The strength of an association between two factors is one of the criteria for causation, but there are other necessary conditions to provide evidence of a causal relationship [12]. Finally, it must be remembered that PFO does not cause stroke, but is part of the causal nexus.

Secondly, RoPE is derived from studies that generally excluded overt venous thromboembolic (VTE) disease. Paradoxical embolism was only a speculation, i.e. without direct evidence of venous or right heart thrombus in the patients’ data used for building the statistical model. Therefore the RoPE score lacks incorporating factors such as deep venous thrombosis, pulmonary embolism, and risk factors for VTE [13].

A group of thought-leading physicians representing the Italian scientific community of cardiology, neurology, and hematology recently proposed a standardized approach on the evaluation and management for patients with PFO and cryptogenic stroke [14]. In this consensus statement recommendations were put forth for diagnostic workup, analysis of probable causation in different patient scenarios, multidisciplinary approaches to diagnosis and treatment, and therapy (both medical and interventional), with additional recommendations made on patient follow-up and device related complications.

An accompanying editorial was skeptical of the validity of the proposed patient management guideline. “Our criticism of this guideline lies less with specific opinions than in the discrepancy between phrases like “must be” in their recommendations and the quality of the evidence to support them” [15]. Indeed, evidence-based rather than opinion-driven, is preferable in the determination of the factors associated with a “PFO-related” stroke and the prediction of recurrence may involve different factors. Unfortunately the evidence base has many gaps of knowledge and conflicting findings. Clinicians and patients need guidance in how to respond to these uncertainties and the Italian contribution is a starting point after acknowledgement of its limitations and the uncertainty of causation in many ischemic strokes.

Three clinical trials, namely CLOSURE, RESPECT, and PC Trial, have been discussed in previous chapters and their findings/conclusions will not be re-iterated here. However, consideration of the following points from these studies should be implemented into the patient evaluation process, individualization of therapeutic strategies, and other lesions learned from trials designed over a decade ago [16–18].

It is useful to consider patient scenarios to better appreciate the diversity of factors involved in decision-making and specifically the variable degree of uncertainty in determining the potential role of a PFO-enabled paradoxical embolism. Table 11.2 lists patient scenarios when PFO closure should be considered due to the presence of clinical findings or special patient circumstances.

Overt venous thromboembolic (VTE) disease at the time of a stroke substantially raises the likelihood of a paradoxical embolic mechanism for the stroke. As presented in the ASCOD classification, it raises the suspicion of causation to a C1 level. In these patients, full anticoagulation and/or PFO closure is often considered despite the lack of randomized clinical trials in this subgroup (Fig. 11.1, Strategies 2,3 and 4). In fact, because of the high index of suspicion in such scenarios, these patients have typically been excluded from randomized clinical trials of PFO closure.

The ASCOD classification acknowledges that grading the likelihood of causation for VTE is not possible due to insufficient workup. This is relevant in terms of including or excluding clinically silent VTE disease by testing. Only rare case studies are published showing the triad of venous thrombosis, right to left shunting and embolism to both pulmonary and systemic circulations [19]. It has been assumed, perhaps incorrectly, that searching for VTE will be fruitless in patients with cryptogenic stroke and PFO. An older study of 42 patients with PFO and “suspected” paradoxical embolism found that 14 % had clinically overt VTE and traditional contrast venography revealed venous thrombi in 57 % [20]. In another study of suspected cardioembolic stroke, 10 % of patients with PFO had deep venous thrombosis with 4 out of 5 having clinically silent VTE [21]. Therefore, the clinician should consider the search for VTE in selected patients since it provides greater certainty for the need for anticoagulation and consideration of PFO closure.

In addition to VTE there are additional clinical scenarios with evidence of right-sided thrombus concurrent with ischemic stroke. These case reports are instructive in highlighting the iatrogenic nature of this subgroup of VTE caused by the proliferation of right-sided pacemaker and defibrillator leads, indwelling central venous catheters, and right-sided prosthetic valves [22–24]. A most intriguing study of 6,075 patients from the Mayo Clinic Rochester demonstrated that the presence of a PFO on routine echocardiography is associated with a substantially increased risk of embolic stroke/TIA in patients with right-sided leads, a presumed nidus for thrombus formation [25]. The potential implications of this study are substantial in studying the role of PFO closure for secondary stroke prevention in this population with a high prevalence of traditional stroke risk factors, overt atherosclerosis, and hypertensive heart disease. Applying the RoPE score to this population would reveal how a low RoPE score would be misleading due to the absence of incorporating potential sources of right-heart thrombus formation in the score.

The lungs with its pulmonary circulation are considered as a filter organ for venous-based embolic debris that is small in size [26]. High-resolution CTA can now detect filling defects in sub-segmental pulmonary arteries as small as 2–3 mm in diameter [27]. In the arena of pulmonary embolism the detection of these small clots may represents over-diagnosis since some evidence indicates that these do not need treatment [27].