Histologic and Immunohistochemical Features of Antemortem Thrombus Compared to Postmortem Clot: Updating the Definition of Lines of Zahn

PMCs were collected from the right atrium or ventricle of the heart in 50 hospital autopsies. Similar to PMCs, perimortem (agonal) thrombi are clinically inconsequential, and so we made no attempt to distinguish them and, for the purpose of this study, included them with PMC. The decedent’s medical records were reviewed and demographic data, including age, sex, postmortem interval between death and autopsy, cause of death, and whether terminal cardiopulmonary resuscitation was performed, were collected. The decedent’s last white blood cell count was recorded as long as it was performed within 24 hours of death. The total weight of the clots and their gross appearances were also documented, with the gross appearances qualified as either chicken fat, red cruor, or a combination of the two. One tissue block per 5g of clot was submitted for microscopy and routinely processed. Sections from each block were stained with hematoxylin-eosin (H&E), CD61 (2f2, Leica Biosystems), and phosphotungstic acid–hematoxylin (PTAH).

The Department of Pathology surgical pathology database at the Cleveland Clinic in Cleveland, Ohio, was queried to identify 50 antemortem arterial thrombi and 50 venous thrombi that had been removed by thrombectomy or embolectomy. The surgical specimens were collected between August 2019 and September 2021. Surgical specimens of AMTs collected from the pulmonary arteries or thrombi “in transit” collected from the right ventricle were categorized as venous. Thrombi that showed organization in the form of infiltrating fibroblasts, collagen deposition, or neovascularization were excluded, as these findings represent unequivocal evidence of AMT. H&E-stained slides were reviewed. CD61 immunohistochemistry was performed to assess platelet distribution, and staining with PTAH was performed to assess the pattern of fibrin deposits. Medical records for each patient were reviewed to determine patient age, sex, and specific location of the thrombus collected.

The following microscopic features of the AMTs and PMCs were documented: (1) The arrangement of fibrin and platelets were grouped into 3 patterns: thick bands formed of nested aggregates of platelets wrapped in fibrin; thin linear or curvilinear strands of fibrin; and amorphous, dense, or loosely clumped fibrin. The patterns of PTAH staining could be grouped into the same categories (Figure 1, A through F). (2) The staining pattern with CD61 was divided into 2 categories: speckled, sparse pattern, or geographic stripes with a speckled background (Figure 2, A and B). (3) Neutrophil karyorrhexis was semiquantified by using a grading scale of 0 through 3, with 0 = absent, 1 = present in a single high-power field (×400 or greater), 2 = present in more than 1 high-power field, and 3 = visible at low power (×200 or less). The sensitivity and specificity of each histologic feature for arterial or venous AMT were calculated.

Autopsies were performed with informed consent from the legal next-of-kin. The institutional review board of the Cleveland Clinic approved this study and institutional guidelines regarding human experimentation were followed.

The 50 autopsies from which PMCs were collected included 26 men (52%) and 24 women (48%) with an average age of 63.1 years. The average weight of the clot collected was 18.96 g per case; the average number of blocks submitted was 4 per case. The average postmortem interval between time of death and time of autopsy was 36.06 hours, with a range of 7 hours to 96 hours. Causes of death were most commonly related to infection (including COVID-19) (20 of 50, 40%), followed by cardiovascular or cerebrovascular disease (10 of 50, 20%), malignancy (8 of 50, 16%), surgical complications (3 of 50, 6%), interstitial lung disease (2 of 50, 4%), and others (7 of 50, 14%). The latter included pulmonary thromboemboli (1 of 50, 2%), gastrointestinal bleeding (1 of 50, 2%), cirrhosis (1 of 50, 2%), long-term complications following dual-organ transplant (1 of 50, 2%), chronic kidney disease (1 of 50, 2%), idiopathic thrombocytopenic purpura (1 of 50, 2%), and cardiac microvesicular steatosis with congenital anomalies (1 of 50, 2%).

The 50 arterial AMTs were collected from 25 men (50%) and 25 women (50%) with an average age of 65.06 years. The locations of the 50 thrombi were most commonly the femoral or popliteal arteries (31, 62%), brachial or radial arteries (11, 22%), iliac artery (4, 8%), superior mesenteric artery (1, 2%), or “other” arterial grafts (3, 6%).

The 50 venous AMTs were collected from 26 men (52%) and 24 women (48%), with an average age of 59.52 years. The most common origins of these 50 thrombi were pulmonary arteries (42, 84%); inferior vena cava, iliac veins, femoral veins (6, 12%); or right atrium (2, 4%).

The gross appearances of the PMC were consistent with their common descriptors of either chicken fat, red cruor, or a mixture of the 2 (Figure 3, A through C). The chicken fat of fresh PMC was a glossy, gelatinous yellow substance; the red cruor was a soft, moist, very dark red substance with a generally smooth surface that took the shape of the cavity from where the clot was retrieved. The PMC was typically not adherent to the right heart endocardium and could be removed in 1 piece except for parts lodged between the trabeculae or pectinate muscles. On cut sections, layering of red blood cells and serum/fibrin material could be appreciated in PMC (rectangular areas in Figure 3, A). In contrast, both arterial and venous thrombi showed granular surfaces, with variable white-tan to red coloration, and were tubular or cylindrical in shape when uncoiled (Figure 3, B and C). AMTs appeared dry and friable. Cut sections of AMTs were more variable and more likely to show coarse, pale gray stripes at the periphery of the thrombus while the core of the thrombus was mostly dark red.

By light microscopy (Figure 4, A and B; Figure 5, A through F), red cruor was composed of aggregated erythrocytes with scattered individual leukocytes admixed with clumps of bone marrow elements and few dispersed platelets (Figure 4, A). The areas of pure chicken fat consisted of a delicate meshwork of fibrin and platelets, with scattered areas of dense aggregated fibrin. Entrapped in the chicken fat clot were few erythrocytes and leukocytes (Figure 4, B). Within the red cruor and at areas of interface between red cruor and chicken fat (Figure 5, A), the 3 patterns of platelet and fibrin deposition were observed (Table 1). Thin curvilinear strands of fibrin that were between 1 to 10 red blood cells in width were seen in 42 of 50 cases (84%), often entirely throughout the red cruor, and amorphous clumps of fibrin were seen in 50 of 50 cases (100%), most commonly seen at the periphery of fragments (Figure 5, C). Rare foci of platelets wrapped in fibrin were identified in only 2 of 50 cases (4%) by H&E staining (Figure 5, E). Neutrophil karyorrhexis in PMC was graded as 0 in 10 of 50 cases (20%), grade 1 in 14 of 50 cases (28%), and grade 2 in 26 of 50 cases (52%). When present, karyorrhexis was more often found at the periphery of the PMC and within bone marrow elements (Figure 6, A through F). No PMC had grade 3 neutrophil karyorrhexis. In the 2 PMC cases showing nested platelets wrapped in fibrin on H&E, PTAH staining confirmed the pattern. It is important to emphasize that the pattern was only present focally in both, and CD61 staining showed a speckled pattern. Neutrophil karyorrhexis in these 2 cases was graded as 2.

By light microscopy, the serpiginous irregular bands of platelets wrapped in fibrin were the most distinguishing features in both arterial and venous AMT, typically present in all submitted fragments and arranged in a concentric or laminated pattern alternating with layers of red blood cells (Figure 5, B and F). Forty-nine of 50 arterial AMTs (98%) and 50 of 50 venous AMTs (100%) showed nested masses of platelets wrapped in fibrin, and this feature had a sensitivity of 99% and a specificity of 96% for AMT (Table 1). The thin fibrin strands and fibrin clumps in AMT were universally identified in red blood cell–rich areas (Figure 5, D). Of the arterial AMTs, 48 of 50 (96%) showed thin curvilinear fibrin strands, and 49 of 50 (98%) showed fibrin clumps. Similarly, 50 of 50 venous AMTs (100%) showed fibrin strands, and 47 of 50 (94%) showed fibrin clumps by light microscopy. Thin fibrin strands had a sensitivity and specificity of 98% and 16%, respectively, and fibrin clumps had a sensitivity and specificity of 96% and 0%, respectively, for AMTs (Table 1). In arterial AMT, neutrophil karyorrhexis was graded as 1 in 7 of 50 cases (14%), grade 2 in 15 of 50 cases (30%), and grade 3 in 28 of 50 cases (56%). In venous AMT, neutrophil karyorrhexis was graded as 1 in 2 of 50 cases (4%), 2 in 7 of 50 cases (14%), and 3 in 41 of 50 cases (82%) (Figure 6, A through F). No arterial or venous AMTs were graded as 0. Overall the sensitivity and specificity, respectively, for each grade of neutrophil karyorrhexis was as follows: grade 0: 0% and 80%; grade 1: 9% and 72%; grade 2: 22% and 48%; grade 3: 69% and 100% (Table 1).

Bone marrow elements, consisting of immature myeloid precursors, nucleated red blood cells, and occasional megakaryocytes, were identified in 38 of 50 PMCs (76%) without accompanying adipose tissue (Figure 7, A and B). All of the decedents had a complete blood count performed within 24 hours of death. The average terminal white blood cell count was 15.43 k/μL for all autopsy patients. For those with bone marrow elements in the PMC, the average terminal white blood cell count was 16.01 k/μL compared to 13.60 k/μL in those without. Records review showed that, of decedents with bone marrow elements in the PMC, 14 of 38 (37%) received terminal cardiopulmonary resuscitation, compared to 3 of 12 decedents (25%) without bone marrow elements. None of the arterial or venous AMTs had bone marrow elements present.

PTAH stain in PMC showed thin fibrin strands in 44 of 50 cases (88%), clumps of fibrin in 50 of 50 cases (100%), and nests of platelets wrapped in fibrin in 2 of 50 cases (4%). PTAH staining in arterial AMT showed thick bands of platelets wrapped in fibrin in 49 of 50 cases (98%), thin fibrin strands in 50 of 50 cases (100%), and fibrin clumps in 47 of 50 cases (94%). Similarly, in venous AMT, PTAH staining showed nests of platelets wrapped by fibrin in 50 of 50 cases (100%), thin fibrin strands in 50 of 50 cases (100%), and fibrin clumps in 49 of 50 cases (98%). The sensitivity and specificity of thin fibrin strands by PTAH staining was 100% and 12%, respectively, and for fibrin clumps by PTAH staining, 97% and 0%, respectively. Nests of platelets wrapped in fibrin seen on PTAH stain had better sensitivity and specificity of 99% and 92%, respectively (Table 1).

CD61 immunohistochemistry in PMC showed a speckled pattern in 47 of 50 (94%) and a geographic pattern in 2 of 50 (4%). One PMC showed essentially no staining, in a patient who died from an intracranial hemorrhage due to idiopathic thrombocytopenic purpura. In arterial AMT, CD61 showed a geographic pattern in 49 of 50 cases (98%), and a speckled pattern in 1 of 50 (2%). In venous AMT, CD61 showed a geographic pattern in 50 of 50 cases (100%). The sensitivity and specificity of a geographic CD61 staining pattern for AMT was 98% and 96%, respectively (Table 1).

The gross appearance of PMC in this study was consistent with its previous descriptions from the literature. PMC of the chicken-fat type is unlikely to be mistaken on gross examination for AMT, and its histologic features are similarly distinctive, being composed largely of fibrin and platelets. The red cruor type of PMC may be more ambiguous on gross examination, and the data confirm the frequent presence of fibrin clumps and thin strands of fibrin in both PMC and AMT. In PMC, these thin strands are seen in nearly every section containing red cruor and can be easily misinterpreted as lines of Zahn.

We found that many of the available descriptions of lines of Zahn lack sufficient detail to be specific for AMT. Based on the common usage of the term, and our findings in this study, we define lines of Zahn as the fibrin and platelet pattern most specific for AMT, consisting of thick, serpiginous bands formed by nests of platelets wrapped in fibrin (Figure 8, A through I). These bands were typically found in multiple submitted fragments, distributed throughout the entire thickness of the thrombus, and were arranged in a concentric or laminated fashion alternating with layers of red blood cells. This pattern was observed in nearly all AMTs. The 1 arterial AMT (procured from the femoral artery) that did not show platelets wrapped in fibrin on H&E also failed to show any on PTAH staining. However, CD61 staining did reveal a focal geographic pattern. On review of this case, the submitted section was composed mostly of red blood cells and it is possible that more thorough sampling would have revealed areas of fibrin and platelets that would show the wrapping pattern. In total, though, nests of platelets wrapped in fibrin are a microscopic feature with high sensitivity and specificity for AMT. We observed a high degree of heterogeneity within the samples, a feature other authors have noticed as well,⁶  which further emphasizes the importance of thorough histologic sampling of thrombi.

There were 2 PMC cases that showed platelets wrapped in fibrin on H&E staining. However, this platelet-fibrin pattern was only focally present and the CD61 staining showed a speckled pattern. The leukocytes around the platelet-fibrin pattern also did not show karyorrhexis. In 2 other PMC cases, CD61 showed a geographic staining pattern; however, it is unlikely that these clots would have been misclassified as thrombi as the pattern comprised a very minor component of the overall sample. In the first case, many of the H&E sections were composed of classic chicken fat morphology; in the second, PTAH staining showed the geographic areas were amorphous clumps of fibrin and platelets and the majority of the specimen was composed of red cruor. Both examples had bone marrow elements present, and neutrophil karyorrhexis was only seen on high power (graded as 1 and 2). These cases illustrate the importance of considering the totality of the specimen, and not just 1 feature, when evaluating the antemortem versus postmortem nature of a possible thrombus.

This is the first study to describe the presence of neutrophil karyorrhexis in PMC, which has been previously cited as definitive for AMT.²  Small foci of neutrophil karyorrhexis could be identified at high power in 80% of PMCs, making this a potentially misleading criterion. However, the karyorrhectic cells are often seen in a background of bone marrow elements (Figure 6, A, C, and E). Widespread neutrophil karyorrhexis on low-power examination was a more specific feature for AMT. This is perhaps not surprising, as neutrophils and neutrophil extracellular traps are known to be a factor in the formation of thrombi.^20,21  Activated platelets and endothelium in the area of the thrombus can continue to chemoattract neutrophils in the presence of flowing blood.^20–22  In PMC, passive activation of platelets and neutrophils in the immediate vicinity of the endothelium may be able to occur at the very end of life. However, without active circulation of blood, chemoattraction is limited, ATP is depleted, and acidosis develops, preventing further neutrophil recruitment and karyorrhexis.

This study shows that evaluation of the leukocytic cellular components of clots and thrombi was also helpful in differentiating PMC from AMT. The leukocytic component of true AMT is predominantly composed of neutrophils, while the presence of bone marrow elements in PMC has not been previously reported. We observed bone marrow elements in 38 PMCs (76%). This phenomenon did not appear to be correlated with cardiopulmonary resuscitation, white blood cell count, or clinical diagnosis. In each case, there was no accompanying adipose tissue, suggesting that they were not the product of embolization from terminal resuscitation. In fact, only 37% of those with bone marrow elements in their PMC underwent cardiopulmonary resuscitation. And although the average terminal white blood cell count was elevated in those with bone marrow in the PMC (16.01 k/μL), many decedents had white blood cell counts within normal range. It may be that this is a finding influenced by multiple factors; the performance of cardiopulmonary resuscitation and elevated white blood cell count may promote circulation of marrow elements. It is also possible that as part of the dying process, chemokines are released that induce peripheralization of marrow cells. It is worth considering scenarios in which it is medically plausible for the decedent to have circulating marrow elements, such as sepsis, severe infection, or hematologic malignancy. In these instances, this finding may be less specific. Nevertheless 76% is an unexpectedly high percentage of PMC cases showing bone marrow elements, making this finding strongly supportive of PMC.

Previous studies have included the histologic and immunohistochemical features of surgically resected thrombi for purposes other than comparison to PMC. The provided photomicrographs from these studies show similar patterns to those described here, on hematoxylin-eosin–stained sections,^23–25  immunohistochemical stains for platelets (CD61, integrin, CD42b),^6,9,20,25,26  and special stains for fibrin.^9,26 

There are potential limitations to this study. The authors could not be blinded in their evaluation of slides by the nature of the specimens, and therefore an element of confirmation bias cannot be entirely excluded. Blinding the pathologist before histologic evaluation may be a valuable way to validate our findings in the future. The extent of sampling was also different between autopsy and surgical specimens. While sampling of PMC was performed by only 1 investigator (A.R.K.), sampling of surgical specimens was performed by pathologists and pathology assistants. This study did not include traumatic or sudden unexpected deaths, although we see no theoretical reason for PMCs in these deaths to differ from those in hospital-based deaths. This study did not address the potential etiologies or clinical significance of PMC, although the authors consider this a topic worthy of future exploration.

Historically, a distinction has been made between “white thrombi” and “red thrombi,” corresponding to those in the arterial and venous systems, respectively^1,8 ; the gross coloration is allegedly based on the relative quantity of fibrin and platelets versus erythrocytes. While we did not formally quantitate the relative composition of arterial and venous thrombi, the data presented here would suggest that distinction between white and red thrombi may be an oversimplification. Lastly, a definite time interval between formation of the AMTs and surgical thromboembolectomy cannot be determined for most surgical specimens, owing to lack of adequate documentation in the electronic medical record and the emergency of the procedure. This time interval could have affected the extent of neutrophil karyorrhexis as evidenced in Table 1.

This study highlights the importance of reevaluating historical descriptions and terms, particularly when there is no clear consensus on the definition of the term. We propose that rather than abandoning the term lines of Zahn, the definition of lines of Zahn can be expanded and updated to include a microscopic feature consisting of thick bands composed of nests of platelets wrapped by fibrin alternating with aggregates of red blood cells. Rarely, lines of Zahn may be observed on gross examination but this feature should be verified by microscopy. Moreover, this feature alone is still not entirely specific for AMT.

We recommend the following microscopic features be evaluated when distinguishing AMT from PMC: (1) platelet-fibrin pattern; (2) predominant leukocytic component of the thrombus; and (3) extent of neutrophil karyorrhexis (Table 2). AMT displays serpiginous bands of nested platelets encased with fibrin (the “true” lines of Zahn) (Figure 8, A through I), which can be verified with a geographic staining pattern of CD61 (Figure 8, C, F, and I), and abundant neutrophils with karyorrhexis apparent at low power (Figure 6, B). Our data show nested platelets wrapped in fibrin to be a highly sensitive and specific microscopic feature of AMT, but the leukocytic component and extent of neutrophil karyorrhexis also proved to be useful features to evaluate. Validating these criteria on thrombi found at autopsy would be a valuable future project. While PMC may show linear strands and clumps of fibrin, the entrapped leukocytes are distinctly polymorphous with easily recognizable bone marrow elements that may contain rare karyorrhectic cells. Staining with PTAH tended to reemphasize the features seen by H&E staining in both PMC and AMT; consequently, it may be helpful to clarify the fibrin pattern for pathologists who do not frequently microscopically evaluate thrombi. CD61 staining is another tool that may help to clarify difficult cases. The gross appearance and clinical history should also be considered. Pathologists must be aware of the potential for morphologic overlap between AMT and PMC and should base their diagnosis on the cumulative weight of all microscopic features.