Can neuroimaging differentiate PFO and AF-related cardioembolic stroke from the other embolic sources? Clinical-radiological correlation on a retrospective study

Development of an MRI based artificial intelligence model for the identification of underlying atrial fibrillation after ischemic stroke: a multicenter proof-of-concept analysis

Background

Atrial fibrillation (AF) represents a major risk factor of ischemic stroke recurrence with serious management implications. However, it often remains undiagnosed due to lack of standard or prolonged cardiac rhythm monitoring. We aim to create a novel end-to-end artificial intelligence (AI) model that uses MRI data to rapidly identify high AF risk in patients who suffer from an acute ischemic stroke.

Methods

This study comprises an internal retrospective cohort and a prospective cohort from Shanghai sixth people's hospital to train and validate an MRI-based AI model. Between January 1, 2018 and December 31, 2021, 510 patients were retrospectively enrolled for algorithm development and performance was measured using fivefold cross-validation. Patients from this trial were registered with http://www.chictr.org.cn, ChiCTR2200056385. Between September 1, 2022 and July 31, 2023, 73 patients were prospectively enrolled for algorithm test. An external cohort of 175 patients from Huashan Hospital, Minhang Hospital, and Shanghai Tenth People's Hospital was also enrolled retrospectively for further model validation. A combined classifier leveraging pre-defined radiomics features and de novo features extracted by convolutional neural network (CNN) was proposed to identify underlying AF in acute ischemic stroke patients. Area under the curve (AUC), sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were calculated for model evaluation.

Findings

The top-performing combined classifier achieved an AUC of 0.94 (95% CI, 0.90-0.98) in the internal retrospective validation group, 0.85 (95% CI, 0.79-0.91) in the external validation group, and 0.87 (95% CI, 0.90-0.98) in the prospective test group. Based on subgroup analysis, the AI model performed well in female patients, patients with NIHSS > 4 or CHA2DS2-VASc ≤ 3, with the AUC of 0.91, 0.94, and 0.90, respectively. More importantly, our proposed model identified all the AF patients that were diagnosed with Holter monitoring during index stroke admission.

Interpretation

Our work suggested a potential association between brain ischemic lesion pattern on MR images and underlying AF. Furthermore, with additional validation, the AI model we developed may serve as a rapid screening tool for AF in clinical practice of stroke units.

Funding

This work was supported by grants from the National Natural Science Foundation of China (NSFC, Grant Number: 81871102 and 82172068); Shanghai Jiao Tong University School of Medicine, Two-Hundred Talent Program as Research Doctor (Grant Number: SBR202204); Municipal Science and Technology Commission Medical Innovation Project of Shanghai, (Grant/Award Number: 20Y11910200); Research Physician Program of Shanghai Shen Kang Hospital Development Center (Grant Number: SHD2022CRD039) to Dr. Dong Huang and the SJTU Trans-med Awards Research (No. 20220101) to Dahong Qian.

Imaging features for the identification of atrial fibrillation in cryptogenic stroke patients

BACKGROUND

Whether specific imaging aspects can be used to identify cryptogenic stroke (CS) patients with high risk of underlying atrial fibrillation (AF) remains unclear. The purpose of this study was to evaluate brain-imaging features in CS patients and their utility as AF predictors.

METHODS

The Nordic Atrial Fibrillation and Stroke study was a prospective observational study of CS and transient ischemic attack patients undergoing 12-month cardiac-rhythm monitoring, biomarker and clinical assessments. In this imaging sub-study, brain magnetic resonance imaging and computed tomography scans from 106 patients were assessed for acute and chronic ischemic lesions in relation to AF occurrence and included in a score to predict AF. Receiver operating characteristics (ROC) curve was used to evaluate the discriminative ability of the score and for its dichotomization for predictive model.

RESULTS

Age, periventricular white-matter hyperintensities (PVWMH), acute lesion size, and vessel occlusion were significantly associated with AF. Acute and chronic cortical infarcts as well as chronic cerebellar infarcts were numerically more frequent in the AF group than the non-AF group. A score consisting of six features (0-6 points) was proposed (age ≥ 65 years, chronic cortical or cerebellar lesions, acute cortical lesions, PVWMH ≥ 2 in Fazekas scale, vessel occlusion, and acute lesion size ≥ 10 mm). Area under ROC curve was 0.735 and a score of ≥ 3 points was a predictor of AF.

CONCLUSIONS

The suggested score was shown to identify CS patients with an increased risk of underlying AF.

In young patients with stroke of undetermined etiology, large vessel occlusions are less frequent in the group with high-risk patent foramen ovale

INTRODUCTION

Patent foramen ovale (PFO) is present in a significant proportion of young patients with stroke of undetermined etiology, but is not always causal. Therefore, classifications (RoPE, PASCAL) have been developed to determine the probability that PFO is the stroke cause. However, the presence of an initial arterial occlusion as a prediction factor was not studied when these classifications were built. Our aim was to evaluate the presence of arterial occlusion in young patients with stroke of undetermined etiology with/without high-risk PFO.

METHODS

From a prospectively-built monocentric database, we identified patients aged≥18 to<60-years with strokes of undetermined etiology and complete etiological work-up, including transesophageal echocardiography. We divided patients in two groups: (i) with high-risk PFO [i.e. PFO with large interatrial shunt (>30 microbubbles) or associated with atrial septal aneurysm] and (ii) with low-risk/without PFO. We recorded the presence of arterial occlusion and large vessel occlusion (LVO) in the acute phase.

RESULTS

We included 96 patients; 55 (57%) had high-risk PFO. Their median age was 48 (40-52) years, and 28 (29%) were women. The percentages of patients with arterial occlusion and with LVO were lower in the high-risk PFO group than in the low-risk/without PFO group: 11 (20%) versus 19 (46%) (P=0.008), and 5 (9%) versus 15 (37%) (P=0.002), respectively. There was no difference in the median RoPE score between groups (P=0.30).

CONCLUSION

The presence of LVO could represent a "red flag" of PFO causality in stroke of undetermined etiology, and could be implemented in future PFO-related stroke classifications.

Differences in clinical and biological factors between patients with PFO-related stroke and patients with PFO and no cerebral vascular events

Background

While stroke is one of the most dissected topics in neurology, the primary prevention of PFO-related stroke in young patients is still an unaddressed subject. We present a study concerning clinical, demographic, and laboratory factors associated with stroke and transient ischemic attack in patients with patent foramen ovale (PFO), as well as comparing PFO-patients with and without cerebrovascular ischemic events (CVEs).

Patients and methods

Consecutive patients with PFO-associated CVEs were included in the study; control group was selected from patients with a PFO and no history of stroke. All participants underwent peripheral routine blood analyses, as well as, on treating physician's recommendations, screening for thrombophilia.

Results

Ninety-five patients with CVEs and 41 controls were included. Females had a significantly lower risk of CVEs than males (p = 0.04). PFO size was similar between patients and controls. Patients with CVEs had more often hypertension (n = 33, 34.7%), p = 0.007. No significant differences were found between the two groups with regard to routine laboratory tests and thrombophilia status. Hypertension and gender were identified in a binomial logistic regression model as independent predictors for CVEs, but with an area under the ROC curve of 0.531, suggesting a very poor level of discrimination between the two groups.

Discussion and conclusions

There is little difference between patients with PFO with and without CVEs in terms of PFO size and routine laboratory analyses. While still a controversial topic in the specialty literature, classic first-level thrombophilic mutations are not a risk factor for stroke in patients with PFO. Hypertension and male gender were identified as factors associated with a higher risk of stroke in the setting of PFO.

Evaluation of right-to-left shunt on contrast-enhanced transcranial Doppler in patent foramen ovale-related cryptogenic stroke: Research based on imaging

BACKGROUND

Cardiogenic embolism caused by patent foramen ovale (PFO) is a common etiology of cryptogenic stroke (CS), particularly in young and middle-aged patients. Studies about right-to-left shunt (RLS) detection using contrast-enhanced transcranial Doppler (c-TCD) are numerous. According to the time phase and number of microbubbles detected on c-TCD, RLS can be classified and graded. We hypothesized that the characteristics of an infarction lesion on diffusion-weighted imaging differs when combining the type and grade of RLS on c-TCD in patients with PFO-related CS.

AIM

To explore the characteristics of infarction lesions on diffusion-weighted imaging when combining the RLS type and grade determined by c-TCD.

METHODS

We retrospectively evaluated CS patients from August 2015 to December 2019 at a tertiary hospital. In total, 111 PFO-related CS patients were divided according to whether RLS was permanent (microbubbles detected both at resting state and after the Valsalva maneuver) or latent (microbubbles detected only after the Valsalva maneuver) on c-TCD. Each group was subdivided into small, mild and large RLS according to the grade of shunt on c-TCD. A normal control group was composed of 33 patients who suffered from simple dizziness. Intragroup and intergroup differences were analyzed in terms of clinical, laboratory and diffusion-weighted imaging lesion characteristics. The correlation between RLS grade evaluated by c-TCD and size of PFO determined by transesophageal echocardiography were also analyzed.

RESULTS

In 111 patients with PFO-related CS, 68 had permanent RLS and 43 had latent RLS. Clinical characteristics and laboratory tests were not significantly different among the permanent RLS, latent RLS and normal control groups. The proportion of patients with multiple territory lesions in the permanent RLS group (50%) was larger than that in the latent RLS group (27.91%; P = 0.021). Posterior circulation was more likely to be affected in the latent RLS group than in the permanent RLS group (30.23% vs 8.82%, P = 0.004). Permanent-large and latent-large RLS were both more likely to be related to multiple (P_trend = 0.017 and 0.009, respectively), small (P_trend = 0.035 and 0.006, respectively) and cortical (P_trend = 0.031 and 0.033, respectively) lesions. The grade of RLS evaluated by c-TCD was correlated to the size of PFO determined by transesophageal echocardiography (r = 0.758, P < 0.001).

CONCLUSION

Distribution of the infarct suggested the possible type of RLS. Multiple, small and cortical infarcts suggest large RLS induced by a large PFO.

Patterns of Infarction on MRI in Patients With Acute Ischemic Stroke and Cardio-Embolism: A Systematic Review and Meta-Analysis

Background: Cardioembolic strokes are common however atrial fibrillation, the most common cause, is often asymptomatic and difficult to detect. There is evidence that infarct topography and volume on magnetic resonance imaging may be associated with specific stroke etiologies.

Aim: A systematic review and meta-analysis were undertaken to summarize the available evidence on the association between stroke etiology, infarct topography, and volume.

Methods: A systematic review was conducted using Medline (OVID), Embase (OVID), and PubMed databases. Hand searches of the gray literature and of reference lists in relevant articles were also performed. A quality assessment was undertaken, based on the STROBE checklist. For each study, the number of patients with and without a CE source of stroke and infarct topography was collected and outcomes presented as odds ratios (OR) with 95% CI and p-values.

Results: Four thousand eight hundred and seventy-three patients with ischemic stroke were included, of whom 1,559 were determined to have a CE source. Bilateral infarcts (OR 3.41; 95% CI 2.20–5.29; p < 0.0001) and multiple territory infarcts (OR 1.57; 95% CI 1.12–2.21; p = 0.009) were more common in patients with a CE source of stroke, than patients without a CE source. Lacunar infarcts (OR 0.49; 95% CI 0.31–0.80; p = 0.004) were more likely to occur in patients without a CE source. No significant difference between the frequency of multiple infarcts (OR 0.96; 95% CI 0.57–1.61; p = 0.87) anterior circulation (OR 1.45; 95% CI 0.83–2.53; p = 0.19) or posterior circulation infarcts (OR 1.06; 95% CI 0.72–1.57; p = 0.75), between the two groups were identified. Three out of four studies examining volume, found a significant association between increased infarct volume and CE source of stroke. A sensitivity analysis with cryptogenic and undetermined stroke sources assumed to be cardioembolic, did not alter the associations observed.

Conclusion: The findings of this systematic review and meta-analysis are broadly consistent with previous literature and provide more robust evidence on the association between infarct topography, volume and stroke etiology. Our findings may assist with refining cardiac investigations for patients with cryptogenic stroke, based on infarct topography.

Clinical RoPE (cRoPE) score predicts patent foramen ovale detection among stroke patients: a multicenter observational study

BACKGROUD

The role of patent foramen ovale (PFO) in cryptogenic stroke (CS) is debated. Tools to predict PFO occurrence and attributable fraction are needed to guide cost-effective diagnostics and treatment. Risk of Paradoxical Embolism (RoPE) score relies on neuroimaging findings, which might be inconclusive in up to 30% of cases.

METHODS

We developed a clinical-based easy tool to predict the presence and attributable fraction of PFO in CS patients, without using neuroimaging. The clinical RoPE (cRoPE) score, ranging 1-10, was elaborated through Delphi method from the original RoPE score, replacing cortical infarction with the Oxfordshire Community Stroke Project (OCSP) classification (lacunar stroke = 0 points, other subtypes = 1 point). Then, from the SISIFO (Studio Italiano di prevalenza nello Stroke Ischemico di pervietà del Forame Ovale, or Prevalence of Patent Foramen Ovale in Ischemic Stroke in Italy) study, a multicenter, prospective study on consecutive acute ischemic stroke patients (n = 1130) classified by Trial of Org 10172 in Acute Stroke Treatment (TOAST) and OCSP criteria and undergoing PFO testing, we selected the VV-CDC cohort (Vibo Valentia, Città di Castello, n = 323) to test the accuracy of cRoPE in predicting PFO detection. We compared cRoPE with RoPE to verify cRoPE reliability. Finally, we tested, through ROC analysis, the performance of cRoPE depending on TOAST classification.

RESULTS

Overall, PFO was detected in 21% in VV-CDC and in 23.4% in remaining SISIFO cohort (n = 807). cRoPE_AUC and RoPE_AUC were similar in VV-CDC. cRoPE performance was comparable with RoPE among CS (cRoPE_AUC 0.76, 95%CI 0.67-0.85, RoPE_AUC 0.75, 95%CI 0.66-0.84). Moving to the remaining SISIFO cohort, cRoPE confirmed satisfactory accuracy in predicting PFO detection in CS patients (cRoPE_AUC 0.71, 95%CI 0.66-0.78, p = 0.032).

CONCLUSIONS

Conclusions: cRoPE might help in stratification of patients with CS, allowing accurate esteem of the likelihood of PFO to be found, especially in cases when neuroimaging is inconclusive.

Patent foramen ovale and neurologic events in patients undergoing liver transplantation

BACKGROUND

Patent foramen ovale (PFO) is present in approximately 20% of individuals. During liver transplantation (LT), intra-operative transesophageal echocardiography can observe transient intra-cardiac shunting of atheromatous debris via a PFO. Closure of PFOs prior to LT has thus been suggested as a potential treatment to reduce peri-operative cerebral vascular accident (CVA). The objective of this study was to assess if the presence of PFO is associated with CVA in patients undergoing LT.

METHODS

Three hundred fifty-eight patients undergoing LT at a single academic institution were included. All patients underwent standardized cardiac evaluation including a detailed cardiovascular history and physical examination, electrocardiogram and transthoracic echocardiogram. Five patients were excluded because of poor transthoracic echocardiographic image quality, and three patients were excluded because of PFO closure prior to LT, yielding a study population of 350 patients. Medical records were reviewed to determine demographics, echocardiographic findings and outcome following LT. Major adverse cardiovascular events, myocardial infarction, CVA and death were collected.

RESULTS

Mean age was 53.4 ± 10.2 years; 61% male and 5% of patients had a prior history of CVA. Alcohol and hepatitis C were the most common etiologies for liver disease. Forty-six patients (13.1%) were diagnosed with PFO prior to LT. In-hospital CVA occurred in 6 patients (1.7%). The prevalence of a CVA was not significantly higher in patients with PFO compared to patients without PFO, 2.2% vs 1.6%, p = 0.57. In-hospital mortality was similar in patients with PFO compared to patients without PFO, 4.4% and 5.3%, p = 1.0.

CONCLUSIONS

The presence of a PFO in patients undergoing LT is not associated with postoperative CVA. Prophylactic closure of PFOs, in the absence of other indications, does not appear to be warranted in patients undergoing LT.