Article Text

Download PDFPDF
Original research
Association between acute ischemic stroke etiology and macroscopic aspect of retrieved clots: is a clot’s color a warning light for underlying pathologies?
  1. Alessandro Sgreccia1,
  2. Zoé Duchmann2,
  3. Jean Philippe Desilles3,
  4. Bertrand Lapergue4,
  5. Julien Labreuche5,
  6. Maeva Kyheng5,
  7. Romain Bourcier2,
  8. Arturo Consoli4
  9. the COMPO-Clot Investigators
  1. 1 Neuroradiology - Interventional Radiology, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy
  2. 2 Department of Diagnostic and Interventional Neuroradiology, Hopital Guillaume et Rene Laennec, Nantes, France
  3. 3 Laboratory of Vascular Translational Science INSERM U1148, Fondation Rothschild, Paris, France
  4. 4 Department of Stroke Centre and Diagnostic and Interventional Neuroradiology, Hospital Foch, University of Versailles and Saint Quentin en Yvelines, Suresnes, France
  5. 5 Santé publique : épidémiologie et qualité des soins, F-59000, CHU Lille, University of Lille EA2694, Lille, France
  1. Correspondence to Dr Alessandro Sgreccia, Radiology, Azienda Ospedaliero Universitaria Ospedali Riuniti di Ancona Umberto I G M Lancisi G Salesi, Senigallia, 60019, Italy; a.sgreccia{at}


Background Few case reports have considered the chromatic aspect of retrieved clots and the possible association with their underlying etiology.

Objective The aim of our study was to analyze the frequency of the TOAST ischemic stroke typical (atrial fibrillation, dissection, atheroma) and atypical (infective endocarditis, cancer-related, valve-related thrombi) etiologies depending on the chromatic aspect of retrieved clots.

Methods A total of 255 anonymized and standardized clot photos of consecutive patients treated by mechanical thrombectomy for acute ischemic stroke were included. A double-blind evaluation was performed by two senior interventional neuroradiologists, who classified the visual aspects of the clots into two main patterns: red/black or white. Main patient characteristics, distribution of underlying stroke etiologies, and outcomes were compared between the two study groups.

Results The inter-reader agreement for clot colors was excellent (k=0.78). Two hundred and thirty-three patients were classified as having red/black clots and 22 as having white clots. A statistically significant association (p=0.001) between atypical etiologies and white clots was observed.

Conclusions White clots were significantly associated with atypical etiologies in this cohort,in particular, with infectious endocarditis.

  • infection
  • stroke
  • thrombectomy
  • tumor
  • intervention

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Although several causes for acute ischemic stroke (AIS) are known, in 20–40% of cases the etiology remains uncertain despite complete diagnostic investigation.1 Some authors recently described some correlations between the histological features of the clots2–5 and the TOAST (Trial of Org 10 172 in Acute Stroke Treatment) categories. Only a few case reports have considered the chromatic aspect of the retrieved clot.6 The aim of this study was to analyze the frequency of cancerous and infectious etiologies and thrombi on prosthetic valves in two specific clot patterns, white or red/black, based on their visual aspect at the moment of withdrawal.


Study population

Two hundred and seventy-seven consecutive patients with large vessel occlusion, treated by mechanical thrombectomy (MT) (with stent-like retrievers or contact aspiration) with or without Intravenous (IV) recombinant tissue plasminogen activator for AIS between May 2017 and July 2018 in two experienced centers and with a photograph of the retrieved clot were included in the study. A total of 277 clot photos were available and 22 were excluded from evaluation of the macroscopic appearance of the clot (colors) because of poor clarity (n=21) or the patient’s age (n=1, 2-year-old child) (online supplemental figure).

Database and etiologies

A prospective and registered multicenter database (ETIS Registry, NCT03776877) was used to export the data on the etiology of the AIS for each patient after a 3-month follow-up. Clinical data, such as age, sex, baseline, and 3-month modified Rankin Scale score, baseline and 24-hour National Institutes of Health Stroke Scale (NIHSS) score were collected for each patient. Other variables such as the occlusion site, the thrombus length on non-invasive imaging (when available), the recombinant tissue plasminogen activator administration, the MT technique, the final Thrombolysis in Cerebral Infarction score and the parenchymal hemorrhage after 24 hours were also collected. An extensive search for the etiology of the AIS was performed. The presence of atrial fibrillation (AF), dissection, and atheroma was verified using normal ECG or ECG Holter when required, carotid Doppler ultrasound, CT angiography/MR angiography, and DSA. Among those patients with an unknown etiology an extensive investigation for cancer and infectious endocarditis (IE) was performed, through a total body CT scan, trans-thoracic (TTE) or trans-esophageal echocardiography (TEE), and blood cultures. We considered that the etiology was cancerous if the following criteria were met: (a) advanced-stage tumors (with evidence of metastases), (b) absence of another possible cause, and (c) evidence of disseminate intravascular coagulation confirmed by biological examinations. All patients with cardiac valves (biologic or mechanic) underwent a TTE or TEE in order to exclude a thrombus on the prosthetic valve.

Retrieved clots and macroscopic features

All the retrieved thrombi (collected from the stent struts, aspiration catheters, or canisters) were immediately rinsed and collected under sterile conditions. The detailed protocol is reported in the online supplemental material. A photograph of each clot positioned on white gauze was taken with a smartphone and all the photos were held in a centralized folder in both centers. All the images were prospectively collected and anonymized with a specific patient-referring number. All the photos underwent a double-blind evaluation performed by two senior interventional neuroradiologists (AC, RB) who classified the clot as ‘white’, or ‘red/dark red’. In cases of discordance between the first two observers a third physician (BL), a vascular neurologist, provided the final evaluation. In three cases a histologic and/or microbiologic analysis of the clot was performed. In the consensus statement by De Meyer et al,2 standardization of the terminology for description of the gross appearance was suggested. The parameters used were the color and the texture of the clot that should be associated with terms such as white/red and homogeneous/heterogeneous, respectively. However, since no specific definitions or guidelines currently exist we proposed to report as ‘white’ a clot which frankly corresponded, in the photos, to that color as well as for the ‘red/black’ (figure 1)

Figure 1

Example of retrieved clots prepared according to the aforementioned protocol. (A) A red/black clot and (B) a white clot before being stored in the refrigerator box.

Statistical analysis

Data analysis was performed on 255 patients with large vessel occlusion and an available clot photo. Categorical variables were expressed as frequencies and percentages, and continuous variables as median (IQR). Normality of distributions was assessed graphically and by the Shapiro-Wilk test. The level of agreement between the two independent evaluations of clot color (red/black vs white) was evaluated using the accuracy rate, and the kappa (κ) coefficient with their 95% confidence intervals. A κ value <0 indicated no agreement, 0–0.20 slight agreement, 0.21–0.40 fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 substantial agreement, and 0.81–1.00 almost perfect agreement. After resolving discordance by a third independent evaluation, patients were divided into two groups by clot color. The main patient characteristics, distribution of underlying stroke causes, and outcomes were compared between the two study groups using the χ2 test (or Fisher’s exact test when the expected cell frequency was <5) for categorical variables, and the Mann–Whitney U test for continuous variables. For 24-hour NIHSS score change, comparison was adjusted for admission NIHSS score using a non-parametric analysis of covariance. Statistical testing was done at the two-tailed α level of 0.05. Data were analyzed using the Statistical Analysis Software package, release 9.4 (SAS Institute, Cary, North Carolina, USA).


The inter-reader agreement for graded clot color (red/black vs white) was excellent (accuracy, 94.9% (95% CI 92.2 to 97.6); κ coefficient, 0.78 (95% CI 0.67 to 0.90)). After resolving discordance by a third independent evaluation, 233 patients were classified with red/black clots (91.4%, 95% CI 87.9 to 94.9) and the 22 remaining patients with white clots (8.6%, 95% CI 5.1 to 12.1). After extensive research for each patient an underlying etiology was found in 210 cases (82.4%). Main patient and treatment characteristics are provided in table 1. Patients with white clots were significantly younger, had a lower admission NIHSS score, and shorter clots than those with red/black ones.

Table 1

Patient characteristics, overall and according to the macroscopic appearance of clots

The details of underlying stroke causes are reported in table 2 according to clot color. We found no significant difference in typical etiologies (AF, atherosclerosis, or dissection), although patients with white clots had a non-significant lower rate of AF than patients with red/black clots (31.8% vs 51.5%; p=0.078). Atypical etiologies (cancer, IE, heart valve thrombus) were significantly more frequent in patients with white clots than in patients with red/black clots (27.3% vs 4.7%; p=0.001). This difference increased when cancer with a progressive form was considered alone (27.3% vs 1.7%, p<0.001). Details of underlying stroke causes in the 22 individual patients with white clots are reported in online supplemental table 1.

We found no significant differences in outcomes with clot color (online supplemental table 2).

Table 2

Underlying causes of stroke according to the macroscopic appearance of clots


Only a few papers, mainly case reports,2 5 6 have analyzed the gross appearance of the retrieved clot to provide a correlation between the macroscopic aspect of the thrombus and a specific etiology. The results, however, were inconsistent. Some authors used terms such as ‘red’ or ‘white’ mostly referring to the clot’s histological features, such as the prevalence either of red blood cells (RBC) or fibrin/platelets or white blood cells, respectively.2–6 Boeckh-Behrens et al 3 and Sporns et al 4 reported large studies that correlate clot histology and stroke causes according to the TOAST classification. The results showed a clear prevalence of fibrin/platelets content in cardioembolic thrombi compared with atherosclerotic thrombi with a major RBC content. Furthermore, histological similarities between the cardioembolic and cryptogenic clots were reported, suggesting that the majority of thrombi of unexplained origin could be cardioembolic. Also, histologic models were proposed by Duffy et al 7 and Gunning et al 8 in order to drive the attention of the scientific community towards a comparison of clot composition and physical characteristics with MT efficacy.

Typical etiologies: cardioembolic, atheromatous, dissection

Our results confirm the findings described by the authors,2 5 6 9 who suggested that cardioembolic clots might be red in color, indicating maybe a major RBC content. However, we observed that although AF-related clots were mostly red/black, those related to IE and implanted valves were either exclusively or more frequently white. On the contrary, our data for the atheromatous and dissection etiology showed a majority of red/black clots more in line with the results obtained by Boeckh-Behrens et al,3 and Sporns et al, 4 which underlined the major RBC content in atheromatous clots.

Infective endocarditis

In IE the incidence of symptomatic embolism can occur in 13–46% of patients.10 Clinically evident cerebrovascular episodes are described in 12–40% of symptomatic embolism cases and mortality ranges from 48% to 80% in patients with neurological complications such as ischemic stroke or hemorrhage in particular.10 In some studies the incidence of ischemic stroke in IE varied up to 17% of patients10 falling after antimicrobial therapy. During the acute phase of IE, after a mechanical endothelial insult, the formation of a thrombus might be induced by inflammatory cells, platelets, and fibrin deposition on valve leaflets. This process could suggest a possible underlying etiology of white clot cardioembolic source. Non-bacterial thrombotic endocarditis (NBTE) has also been reported to be of cardioembolic source,11 with a high risk of stroke. Valvular vegetations have been described as more fragile than their infective equivalents and more prone to peripheral embolization, which is reported to occur in 50% of patients, associated with a detection rate of only 18% at the TEE.11 12 Furthermore, NBTE is frequently associated with advanced neoplasm.12 13 As reported by several authors clots retrieved in patients with IE showed this histological pattern and a white macroscopical aspect.6 In our cohort, a white clot was observed in all four patients with an underlying endocarditis and the mean age was 42 years. Streptococcus gondonii and Streptococcus oralis mitis were identified in blood cultures, while in one case Candida parapsilosis was found through polymerase chain reaction performed on the clot. The fourth patient had an evolutive cancer and a NBTE was diagnosed. Although endocarditis is a rare clinical condition in acute ischemic stroke, 1.6% in our study population, these results could mean that particular attention should be paid to patients with white retrieved clots, especially in younger patients.

Clots associated with prosthetic valves

Prosthetic valves have been described as a source of embolism.14 Dangas et al 14 reported an annual rate of mechanical valve thrombosis of 0.1–5.7%. In our cohort we observed only two patients with a mechanical valve; in one case we retrieved a red/white clot and in the other we observed a white one. The only patient with a biological valve had no valvular thrombi.


Ischemic stroke in patients with cancer has been reported and, according to the literature, this might be caused by direct tumor effects or by a hypercoagulability state such as Trusseau’s syndrome or hematologic disease.12 13 The most common related lesions are mucin-producing tumors like pancreatic, colon, breast, lung, prostatic, and ovarian adenocarcinomas, which are frequently linked with procoagulating disorders.12 13 Some authors10 12–14 have described these ‘valvular’ and ‘tumorous’ clots as white-colored at macroscopical aspects and rich in fibrin content at histological analysis.

In our cohort only three patients in the red/black group (3/233, 1.3%) and two (2/22, 9.1%) with white clots were found to have an evolutive cancer.

These results could suggest that red/black clots are substantially associated with AF, atheroma, or dissection and, therefore, red/black clots retrieved in cryptogenic strokes could be more likely to be related to typical etiologies. On the contrary, pure white clots should provide more solid suspects on IE, although more than half of white clots were associated with typical etiologies. Although the number of patients was limited to describe an association between white clots and neoplastic syndromes or prosthetic valve thrombi, the rate of white clots was not negligible in these subgroups.


We acknowledge some limitations, mainly comparable to those of any microscopic or macroscopic analysis of clots retrieved from patients with AIS with endovascular devices. First, fibrin-rich (white) thrombi are very difficult to retrieve and consequently, since we analyzed only those we retrieved, fibrin-rich thrombi may be under-represented in our thrombi collection. Second, the use of IV recombinant tissue plasminogen activator before the endovascular therapy can modify the thrombus characteristics. Furthermore, thrombi retrieved after significant chemical and mechanical manipulations could have vastly dissimilar composition and length from the original thrombus. Moreover, considering the relatively small sample of white clots we cannot discuss the diagnostic accuracy. Lastly, there was a lack of a glossary for the classification of the clots since this type of macroscopic analysis has not yet been performed.


Atypical etiologies are significantly correlated with white retrieved clots—in particular, infective endocarditis. In cases of a cryptogenic stroke, an attentive observation of the macroscopic aspect of the clot could influence and drive the diagnostic process and, consequently, the treatment.


We thank Polly Gobin for the English language editing.



  • Contributors All the authors participated with substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work. All the authors participated drafting the work or revising it critically for important intellectual content, they gave the final approval of the version to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

  • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All the data are presented in the manuscript and in the supplemental material and are available upon reasonable request to the corresponding author.

  • Patient consent for publication Not required.