Where Are Clots in Atrial Fibrillation? Did We Have the Wrong Assumptions Over the Last Decades?

Alberto Cresti; Horst Sievert; Ugo Limbruno; Pasquale Baratta; Marco Solari; Francesco De Sensi; Raffaele De Caterina


Europace. 2020;22(6):845-846. 

It is commonly believed that in non-valvular atrial fibrillation (AF) 90% of clots occur in the left atrial appendage (LAA), however, there are few publications cited to support this evidence, and derived from old data.

An EHRA/EAPCI expert consensus statement on catheter-based left atrial appendage occlusion—an update was published on August 2019 in Europace.[1] The document is a complete 'state-of-the-art' paper, useful for clinicians, imagers, and interventional cardiologists.

The availability of occlusion devices and the continuing technological improvements indeed required a thorough update, and the authors' effort in reviewing the huge recent literature on the topic is commendable.

In the first chapter of this consensus document, the background and pathophysiology of thrombus formation in the left atrium are reviewed. Among data supporting the rationale for LAA closure in AF, two publications are cited: in 'non-valvular' AF, the paper by Blackshear and Odell,[2] who described 91% thrombi in the LAA and 9% out of it; and the paper by Mahajan et al.,[3] who described 89% of thrombi in the LAA and 11% outside.

Despite the presence of 9–11% of thrombosis out of the LAA may raise concern about the real efficacy of LAA closure, these two reviews are based on old studies. We believe it is time to have a closer look at them and to frame these data in the more updated literature: today's anticoagulation regimens are largely different from the older ones based on vitamin K antagonists, and cases of rheumatic AF are now a minority in developed countries.

Blackshear and Odell[2] published in 1996 a review dividing patients with rheumatic and non-rheumatic AF, which is quite different from 'valvular' and 'non-valvular' AF, although what is a 'valvular' AF is still currently disputed.[4] Thrombi were found to be located or extending into the left atrial cavity in 57% of patients with rheumatic AF, whereas 91% (201 out of 1288) patients of non-rheumatic AF left atrial thrombi were confined to the LAA. A careful reading of the original papers cited in Blackshear's review raises some further concern: Stoddard et al.[5] described one single case of left atrial thrombus in a patient with a mitral prosthesis out of 317 patients; five cases of right atrial thrombi were also described but a pulmonary embolism was not reported in any individuals.

Manning et al.[2] described one case, out of 233 patients, with a thrombus in the left atrial cavity and six right atrial thrombi in a mixed population of 'valvular' and 'non-valvular' AF.

The case of left atrial thrombus described by Brown et al.[2] was inside the LAA, but protruding in the atrial chamber.

Klein et al.[2] found one single case of thrombus out of LAA, but in this case located in the right atrial appendage.

The two patients with a left atrial cavity thrombus described by Tsai et al.[2] were affected by mitral stenosis.

A second paper by Manning et al.[2] was reviewed by Blackshear in a population of 231 patients undergoing either elective mitral valve surgery or excision of a left atrial tumour: here three cases of left atrial cavity thrombi are reported but, looking at the figures published inside the paper, it is clear that one of them was inside the LAA, while another one was an in-transit venous embolus entrapped in the patent foramen ovalis, an entity that was not yet well studied in 1995.

Finally, 12 out of 21 left atrial cavity thrombi included in the calculations by Blackshear et al. derived from a retrospective, autoptic study published in 1969 by Aberg et al.[6] in a mixed population of patients affected by valvular heart disease, and including cases of pulmonary embolism. Such autoptic 'clots' should not be put together with the thrombi described in vivo during a transoesophageal study and should be kept separated from the other in vivo cases of left atrial thrombi.

Moving to Mahajan's review published in 2012,[3] the authors reported an LAA thrombus in 56% of patients affected by 'valvular' AF, and 11% in 'non-valvular' AF. Looking at Table 3 of this review, the most numerous cases of left atrial thrombi were reported by the same authors previously reviewed by Blackshear: Aberg et al.,[6] Brown et al., Tsai et al., Stoddard et al., Klein et al.[2] Other cases of left atrial cavity clots are reported by Collins et al.[7] in a study published in 1994 in a population of non-anticoagulated patients with non-rheumatic AF: in this old study, one may find that, among four extra-LAA thrombi, the location was in the right atrial appendage, in the right atrial cavity, and two in the left atrial cavity, but in one of these cases the thrombus protruded from the LAA.

Among the most recent studies analysed by Mahajan in 'non-valvular' AF, only one more case was described by Bernhardt et al.,[8] but, again, in this patient, an LAA thrombus protruding from the LAA into the left atrium was present, therefore primarily formed inside the LAA. Table 1 the results of these studies are summarized.

Therefore, from more recent studies in 'non-valvular' AF populations, we should expect a very low frequency of left atrial cavity thrombi. This is confirmed by our large-scale study,[9] of a population of 1420 patients affected by 'non-valvular' AF as we found that all patients (87) had a thrombus located inside the LAA and only one case had a thrombus even in the left atrial cavity, whereas three cases of right atrial appendage thrombi were also detected.

The problem of frequency and clinical significance of right atrial thrombi has previously been discussed by our group,[10] i.e. they appeared to be about 12 times less frequent than in the left atrium/LAA besides, a pulmonary embolism was never described in such cases.

In conclusion, the frequency of left atrial cavity thrombi is most probably largely overestimated if based on older literature and is mostly confined to cases of rheumatic AF. The very low real frequency of left atrial thrombi out of the LAA in more modern reports might be an important rationale for LAA occlusion interventions.