(Stroke. 1998;29:2665-2666.)
© 1998 American Heart Association, Inc.
Letters to the Editor |
Electrocardiographic Diagnosis of Patent Foramen Ovale Associated With Ischemic Stroke
Section of Neurology, Hospital Universitari de Girona Doctor Josep Trueta, Girona, Spain
To the Editor:
The report of Ay et al1 showed the value of an electrocardiographic finding called "crochetage" to identify stroke patients with patent foramen ovale (PFO). In this series of 60 patients with cryptogenic stroke, the sensitivity and specificity of the crochetage pattern for diagnosis of PFO were 36% and 91%, respectively. Nevertheless, as the authors admitted, this study was limited by the very restrictive selection criteria and the low sensitivity of the color transthoracic echocardiography (TTE) used to detect PFO in more than 20% of patients. Additionally, the investigators were not blinded to the crochetage pattern when they selected the study population from those patients with cryptogenic stroke.
Our purpose was to duplicate their results in a nonselected population of patients in whom PFO was evaluated prospectively with a more sensitive technique. Moreover, we analyzed the relationship between the degree of the right-to-left shunt (RLS) and the presence of crochetage.
Between February 1996 and May 1997, we performed a contrast transcranial Doppler (TCD) study in 208 patients admitted consecutively within the first 48 hours of an acute ischemic stroke or transient ischemic attack. The procedure to diagnose and quantify PFO as well as the main results have been published recently.2 Briefly, patients were divided into 3 different groups based on the maximum number of microbubble signals detected in MCA in any single frame after intravenous air-saline contrast injection during Valsalva maneuver: "normal" TCD study (if 0 signals were detected), "small" RLS (<10 signals) and "large" degree of shunt (>10 signals). In this last group, "shower" (>50 microbubbles) and "curtain" (uncountable microbubbles) patterns were considered. Patients with cryptogenic stroke were classified according to predefined criteria that did not include ECG findings other than the absence of flutter or atrial fibrillation.2
We studied all the 68 patients with cardiac RLS and 68 controls randomly selected from the 135 patients without RLS, matched by sex and age. Mean age was 64.40±13.4 years for the case patients and 64.45±12.19 years for the control subjects. There were 48 men and 20 women in both the case and control groups. In all patients we evaluated the ECG performed at admission. Two different physicians blinded to diagnosis independently reviewed the ECG of every patient looking for the crochetage pattern in one or more limb leads (II, III or aVF), defined as a notch in R wave with a rapid up-and-down in the ascending branch or near the zenith which produce a M-shaped or bifid form and always involving the initial 80 ms of the QRS complex.3
The crochetage pattern was identified in 7 patients by the first
physician and in 12 patients by the second (
=0.718). It is
remarkable that no patient classified as having crochetage by the first
examiner was classified by the second in a different way. After
interobserver agreement still blind to the study groups, crochetage was
noted in 10 patients, 2 with cardiac RLS and 8 without
(P=0.049). When patients were classified according to the
presence of a massive RLS (ie, cases with shower or curtain pattern in
TCD), crochetage was not present in patients with massive RLS and
was detected in 10 without a high degree of shunting
(P=0.16) (see the Table
).
We
obtained similar results in the subgroup of 44 patients with
cryptogenic stroke (crochetage was not present in patients with
RLS; P=0.027). These results indicate not only that
crochetage is not related to the presence of PFO but suggest a trend to
a negative relationship.
|
In our series, crochetage could be easily recognized in ECG with small interobserver differences; however, it was not related to RLS nor to its magnitude. Therefore, in our opinion, crochetage is not a useful sign to identify stroke patients with PFO. Several reasons may be argued to explain the opposing results in the article of Ay et al and in our study: first, the conservative selection scheme in the former could lead to a selection bias; second, our study was more sensitive in detecting PFO and used the same diagnostic tool in all the patients, in contrast to the investigation of Ay et al; third, the time between stroke onset and ECG, which was more prolonged in the first study, could modify the ECG tracing. In fact, we can speculate that in some cases crochetage might be due to hemodynamic disturbances or cardiac conduction defects seen particularly in the early acute phase of stroke. We can ruled out a misdiagnosis of patients with PFO because we proved in our series the accuracy of contrast TCD for the diagnosis of PFO in comparison with transesophageal echocardiography.2
In conclusion, the crochetage pattern is not an accurate sign for the diagnosis of cardiac RLS in the acute clinical setting. We think contrast TCD is the ideal method to perform a simple and safe bedside diagnosis of PFO in patients with acute stroke.
References
1.
Ay H, Buonanno FS, Abraham SA, Kistler JP, Koroshetz
WJ. An electrocardiographic criterion for diagnosis of patent
foramen ovale associated with ischemic stroke.
Stroke. 1998;29:1393–1397.
2.
Serena J, Segura T, Pérez-Ayuso MJ, Bassaganyas J,
Molins A, Dávalos A. The need to quantify right-to-left
shunt in acute ischemic stroke: a case-control study.
Stroke. 1998;29:1322–1328.
3. Heller J, Hagège AA, Besse B, Desnos M, Marie FN, Guerot C. "Crochetage" (notch) on R wave in inferior limb leads: a new independent electrocardiographic sign of atrial septal defect. J Am Coll Cardiol. 1996;27:877–882.[Abstract]
Response
Massachusetts General Hospital, Department of Neurology, Stroke Service
Massachusetts General Hospital, Department of Cardiology, Boston, Massachusetts
We thank Tembl et al for their interest in our report, and thank the Editor for the opportunity to provide the following clarifications.
In all scientific work, one must first establish a program, then aim to remove bias, and establish that comparisons may be appropriately drawn. Bias might be avoided via blinding; in fact, we were blinded both to the ECG tracings and to the study groups at the time of patient selection.1 Patients with other ECG abnormalities that might interfere with crochetage were fastidiously excluded a priori on the basis of cardiology reports of their ECGs. Crochetage is not a formally reported pattern at our hospital, and blinding is maintained.
Restrictive patient selection criteria (which led to our small sample size) were important to prevent contamination of the ECGs by those of other cardiac conditions (eg, supraventricular and ventricular arrhythmias, intraventricular conduction abnormalities, myocardial infarction). Otherwise, to identify the influence of these cardiac conditions on crochetage, a much larger sample size would have been needed (as demonstrated by the study of Heller et al,2 with 1560 patients). We studied 60 patients with cryptogenic stroke, 28 of whom had a patent foramen ovale. On the other hand, Tembl et al studied 68 patients with right-to-left shunts, an etiologically heterogeneous group including various heart diseases. Our criteria for the diagnosis of cryptogenic stroke were more strict. Tembl et al did not use MRI and MR angiography to diagnose the stroke type or responsible intracranial stenosis in the 44 patients who were presumed to have cryptogenic stroke. They had a cryptogenic stroke rate of 26%, compared with 11% in our study. Hence, it is difficult, if not impossible, to directly compare their results with ours in respect to crochetage. In their letter, as in the article of Serena et al,3 the authors interchange the terms "patent foramen ovale" and "right-to-left shunting" on transcranial Doppler ultrasonography, clouding the issue regarding the underlying pathology. Eventually, we learn that transesophageal echocardiography was performed in 44 patients with cryptogenic stroke, of whom only 22 had patent foramen ovale.
Despite the methodological discrepancies between the study of Tembl et al and our study, it is still difficult to explain the differences between the studies in observed crochetage. We showed a 36% prevalence rate of crochetage in our patients with patent foramen ovale. Heller et al2 reported crochetage in 66% to 97% of 532 patients with various types of ostium secundum atrial septal defects. The high variability in the prevalence necessitates a clear reassessment of the definition criteria for crochetage. We have frequently observed humplike, subtle excursions on the ascending limb of the R wave; although the pattern on the R wave still appeared M-shaped, the intersections between the arms of the M had a rounded appearance. In our patients, this pattern was not persistent but changed from one R wave to another. The interexaminer concordance rate for these subtle excursions was very low. We did not count them as crochetage. On the other hand, M-shaped patterns with spiky arms were invariably observed on all R waves in a given ECG, with excellent interexaminer concordance. We would also like to point that when the arms of M-shaped patterns are very close to each other, a crochetage may easily be missed, especially if it is situated at the top of R wave; unless special attention is given, the two arms of the M may be perceived as a single upstroke.
Both the sensitivity and positive predictive value increased in our study when patients were excluded who had undergone low-yield echocardiography (color transesophageal echocardiography, used in 16% of all patients, 1 in the patent foramen ovale and 9 in the control groups).
It is unlikely that crochetage is due to the effect of stroke as suggested by Tembl et al, because both the control and patent foramen ovale groups comprised patients with stroke.
The possible relationship between crochetage and degree of right-to-left shunt awaits study in a larger prospective series, in which there are careful definitions of the ECG patterns, the underlying cardiac anatomy, and the degree of interatrial shunting.
Finally, we would like to state that the ECG finding of crochetage does not replace transcranial Doppler ultrasonography or echocardiography for diagnosis of patent foramen ovale. Management of patients with patent foramen ovale is a multistep procedure. First, an initial suspicion may be raised in the emergency setting; ECG can be helpful at this step. Second, demonstration of cardiac right-to-left shunting is necessary to determine the stroke mechanism and to guide acute treatment; contrast transcranial Doppler ultrasonography has proved its value at this level. Third, the morphological characteristics underlying the intracardiac right-to-left shunt (eg, the type of defect, size of the defect, associated atrial septal aneurysm) should be determined before long-term decisions are made regarding platelet antiaggregation, anticoagulation, or possible surgery; a transesophageal echocardiography study remains essential at this step.
References
1. Ay H, Buonanno FS, Abraham SA, Kistler JP, Koroshetz WJ. An electrocardiographic criterion for diagnosis of patent foramen ovale associated with ischemic stroke. Stroke. 1998;29:1393–1397.
2. Heller J, Hagège AA, Besse B, Desnos M, Marie FN, Guerot C. "Crochetage" (notch) on R wave in inferior limb leads: a new independent electrocardiographic sign of atrial septal defect. J Am Coll Cardiol. 1996;27:877–882.
3. Serena J, Segura T, Pérez-Ayuso MJ, Bassaganyas J, Molins A, Dávalos A. The need to quantify right-to-left shunt in acute ischemic stroke: a case-control study. Stroke. 1998;29:1322–1328.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||