The epidemiological data assembled in the review are striking. Active cancer is present in roughly 5 to 10 percent of patients presenting with ischemic stroke, a proportion that rises among those with embolic stroke of undetermined source, the cryptogenic category in which no conventional mechanism can be identified. In 3 to 5 percent of such patients, the stroke actually precedes the cancer diagnosis by up to two years, making it the first clinical manifestation of an occult malignancy. Occult cancer is detected in about 5.3 percent of patients with embolic stroke of undetermined source, with the highest detection rate occurring in the early period after the stroke, at 14.3 per 1,000 person-months. Among cancer patients who do stroke, cancer-associated coagulopathy is the most frequent cause, accounting for 39.8 percent of cases, while traditional mechanisms such as large-artery atherosclerosis and cardioembolism from atrial fibrillation are comparatively less common.

Timing emerges as one of the most consistent signatures of the condition. The risk of hypercoagulability-related stroke peaks within the first six months after a cancer diagnosis, and many patients have already reached metastatic disease by the time the stroke occurs. Arterial thromboembolism more broadly shows a characteristic temporal pattern, with incidence peaking in the first year after cancer diagnosis, and stroke represents the majority of these arterial events at 71.6 percent. Although the risk declines after the first year, it remains elevated above that of the general population for as long as a decade. Histologically, adenocarcinoma dominates, with lung and pancreatic cancer most frequently implicated, followed by colorectal, breast and prostate malignancies. The interval between cancer diagnosis and stroke is particularly short in pancreatic and colorectal cancer, underscoring their especially potent thrombotic potential.

The pathophysiology described in the review is a layered interplay of tumor biology and hemostasis. Adenocarcinomas produce mucin, a heavily glycosylated molecule that, although largely cleared by the liver, can engage the adhesion molecules P-selectin and L-selectin to trigger the formation of platelet-rich microthrombi. Tumor cells also overexpress tissue factor, a transmembrane receptor that activates the coagulation cascade through the extrinsic pathway, and this expression is upregulated by oncogenic events such as inactivation of the p53 tumor suppressor and activating mutations in the K-ras oncogene. More recently, alterations in K-ras and the STK11 tumor suppressor gene have been linked to arterial thromboembolism risk, with hazard ratios of 2.22 and 3.48 respectively for stroke among individuals carrying versus lacking these alterations. A recent in vitro study in human pancreatic cancer cell lines further showed that thrombosis induced by tissue factor-expressing microvesicles depends predominantly on coagulation factors VIII and IX, with von Willebrand factor playing a lesser role.

Beyond mucins and tissue factor, the review highlights two additional mechanisms that have reshaped understanding of cancer-driven thrombosis. Circulating extracellular vesicles secreted by tumor cells have been measured by flow cytometry, and levels of cancer cell-derived vesicles are higher in patients with cancer-related stroke than in other groups, correlating with D-dimer levels but not with vesicle-associated tissue factor, which points to prothrombotic pathways beyond the canonical tissue factor route. In vitro experiments also found that vesicles from adenocarcinoma cell lines shorten clotting times more than those from squamous cell carcinoma, consistent with the clinical predominance of adenocarcinoma. Separately, neutrophil extracellular trap formation, or NETosis, has emerged as a key mediator: cancers sensitize peripheral neutrophils toward NET release via granulocyte colony-stimulating factor, and in stroke patients with cancer, NET-specific biomarkers such as plasma DNA, nucleosomes and citrullinated histone H3 are significantly elevated and correlate with D-dimer and thrombin-antithrombin complex levels. A post-mortem histopathological study even demonstrated widespread microvascular thrombosis in the brain, with H3Cit-positive cells embedded in multiple cerebral thrombi.

Clots physically retrieved from patients during thrombectomy provide direct confirmation of this biology. Thrombi from stroke patients with active cancer are platelet-rich and erythrocyte-poor, a composition most pronounced in those with non-bacterial thrombotic endocarditis, and immunohistochemical analysis shows significantly higher content of von Willebrand factor and citrullinated histone H3 in clots from patients with cancer. Proteomic studies have added further markers, identifying elevated levels of proteins associated with active cancer and immune responses, such as IGHG1 and vitronectin, in cancer-related clots. This distinctive composition has a practical radiological consequence: the absence of the susceptibility vessel sign on susceptibility-weighted imaging in patients with large-vessel occlusion is associated with active cancer, with a reported sensitivity of 27 percent and specificity of 85 percent, reflecting the fibrin- and platelet-heavy, red-cell-poor nature of the thrombus.

On brain imaging, the review identifies a pattern that should raise immediate suspicion. The hallmark of cancer-associated coagulopathy stroke is multiple ischemic lesions spanning more than two vascular territories on diffusion-weighted MRI, with microembolic dispersion seen in 78 percent of patients and prior silent infarcts in half. The most specific finding is the so-called three territories sign, defined by simultaneous bilateral lesions in three vascular territories spanning both anterior and posterior circulation. In the absence of infective endocarditis, this sign is highly specific for cancer-related stroke, and among patients with three-territory infarcts and no identifiable embolic source, approximately 75 percent of cases are malignancy-related. The number of affected territories independently predicts occult malignancy in cryptogenic stroke. Cortical and subcortical regions are most frequently affected, followed by the cerebellar hemispheres and corpus callosum, while deep structures and the brainstem are rarely involved.

Biomarker evidence converges on D-dimer as the single most useful test, despite its lack of specificity. Proposed diagnostic thresholds vary, with one study suggesting a cutoff of 5.5 micrograms per milliliter yielding 99.7 percent specificity and 92.9 percent positive predictive value, and a later study proposing 2.785 micrograms per milliliter with 50.9 percent sensitivity and 98.5 percent specificity. The authors caution, however, that different analytical platforms and reporting units, fibrinogen-equivalent versus D-dimer units, preclude a universal cutoff. D-dimer remains the only biomarker consistently associated with recurrent stroke and mortality in this population, and elevated levels after anticoagulant therapy predict early recurrence. Supporting markers include C-reactive protein above 20 milligrams per liter and fibrinogen above 600 milligrams per deciliter, both highly specific for occult malignancy in cryptogenic stroke, along with CA-125, elevated neutrophil-to-lymphocyte ratio above 15, and transcranial Doppler microembolic signals, though the latter show limited specificity. The NORSTROKE score, combining age, D-dimer, hemoglobin and smoking status, offers a probabilistic approach, estimating a 53 percent probability of active cancer in a patient under 75 with cryptogenic stroke, D-dimer above 3 milligrams per liter, hemoglobin below 12 grams per deciliter and a smoking history.

Treatment remains the weakest link in the evidence chain. Low-molecular-weight heparin and direct oral anticoagulants are the most commonly used secondary prevention strategies, with low-molecular-weight heparin generally preferred in gastrointestinal malignancies because of the higher bleeding risk of direct oral anticoagulants in that setting, while vitamin K antagonists are less effective. Comparative studies have shown broadly similar outcomes between agents: enoxaparin versus aspirin, direct oral anticoagulants versus enoxaparin, apixaban versus aspirin in a post hoc analysis of the ARCADIA trial, and oral edoxaban versus subcutaneous enoxaparin all failed to show significant differences in major bleeding, thromboembolism or survival. A recent retrospective study found no mortality or recurrence difference between anticoagulant and antiplatelet therapy, and a 2026 American Heart Association scientific statement concluded there is insufficient evidence to choose between anticoagulation and antiplatelet therapy for this indication. Current guidelines for cancer-associated thrombosis recommend at least three to six months of anticoagulation, extended when cancer remains active.

The prognosis statistics explain why the authors frame this as an urgent, underrecognized problem. Thirty-day mortality ranges from 25 to 50 percent, compared with 14 percent in stroke patients without cancer, and more than half of deaths occur within six months. The median modified Rankin Scale score at discharge is 3, and at three months 75.6 percent of patients remain at least moderately disabled. Recurrence rates reach 13.6 percent, roughly three times that of patients without cancer, with cumulative rates of 7 percent at one month and 16 percent at six months. When stroke is the first manifestation of occult malignancy, median survival is only 58 days, and median survival in cancer patients with cryptogenic stroke ranges from 62 to 365 days, versus 590 days for stroke of other causes. In patients with non-bacterial thrombotic endocarditis, up to 90 percent die or suffer recurrent stroke within six months. The review’s authors acknowledge limitations, including the predominance of retrospective observational data, the absence of prospective protocol registration, marked heterogeneity that prevented meta-analysis, and language restrictions to English and Spanish. They call for prospective multicenter cohorts, standardized D-dimer assays, external validation of the NORSTROKE score and randomized trials comparing low-molecular-weight heparin, direct oral anticoagulants and antiplatelet therapy. As cancer incidence rises and survival improves, they warn, the burden of this devastating stroke subtype will only grow, making earlier recognition and personalized antithrombotic strategies a clinical imperative.

Subject of Research: Ischemic stroke caused by cancer-associated coagulopathy, including its epidemiology, mechanisms, biomarkers, imaging features and treatment

Article Title: Stroke secondary to cancer-associated coagulopathy: a systematic review

Article References: Stroke secondary to cancer-associated coagulopathy: a systematic review. (n.d.). https://doi.org/10.1007/s00415-026-14119-y

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14119-y

Keywords: stroke, cancer-associated coagulopathy, Trousseau syndrome, adenocarcinoma, D-dimer, hypercoagulability, three territories sign, anticoagulation, low-molecular-weight heparin, direct oral anticoagulants, neutrophil extracellular traps, tissue factor

The study, led by María Payá, Gemma Serrano-Heras, and Tomás Segura of the General University Hospital of Albacete and collaborators across Spain, compared cerebral thrombi from six pre-Omicron COVID-19 stroke patients with those from six matched non-infected controls treated at the same institutions. All patients had experienced acute ischemic stroke due to large vessel occlusion and underwent endovascular thrombectomy, during which the retrieved clot material was carefully preserved for research. Although the cohort was small, the breadth of analytical techniques applied to each thrombus was unusual: whole-slide digital imaging quantified immune cells, platelets, and red blood cells; transmission electron microscopy revealed clot architecture at nanometer resolution; liquid chromatography–tandem mass spectrometry with SWATH acquisition profiled more than 1,200 proteins; and total reflection X-ray fluorescence measured fifteen chemical elements.

Clinically, the contrast between the two groups was sobering. Twenty-four hours after treatment, COVID-19 patients had significantly worse neurological scores than controls, and at ninety days their functional outcomes, measured by the modified Rankin Scale, were markedly poorer. Cryptogenic stroke, in which no clear cause is identified, was the most common subtype in both groups. Routine blood counts showed no overt anemia or thrombocytopenia, but COVID-19 patients tended toward higher monocyte counts, elevated hemolysis indices, increased ferritin, and lactate dehydrogenase levels above the normal range, subtle laboratory signals hinting at red blood cell injury and iron release that had gone largely unrecognized at stroke presentation.

Under the microscope, the COVID-19 clots looked fundamentally different. Immunostaining for citrullinated histone H3 and myeloperoxidase showed similar burdens of neutrophil extracellular traps and neutrophils in both groups, but COVID-19 thrombi contained a trend toward more macrophages displaying a diffuse, scattered CD68 staining pattern suggestive of a hyperactivated state. Platelet content, quantified by CD61 staining, was reduced by roughly a third in the infected clots. Histology revealed an extracellular fibrillar matrix that was irregular and less densely packed, and electron microscopy confirmed a disordered ultrastructure: instead of the well-organized peripheral fibrin meshwork enclosing packed cellular regions seen in control clots, COVID-19 thrombi showed fibrillar material strewn chaotically through the section, along with numerous membrane protrusions resembling the filopodia of highly activated platelets.

The most striking discovery emerged from elemental and proteomic analysis. The COVID-19 clots had a distinctive dark-red, brownish coloration that persisted even in protein extracts, a visual clue typically pointing to heme pigments. Total reflection X-ray fluorescence confirmed markedly elevated iron concentrations: more than 80 percent of the COVID-19 thrombi exceeded nine sulfur-normalized units of iron, compared with fewer than 30 percent of controls, while all other measured elements were unchanged. Crucially, the excess iron could not be explained by a simple excess of red blood cells, because glycophorin A immunostaining showed comparable erythrocyte content in both groups.

Proteomics then revealed where that iron was coming from. Of 720 proteins shared between the two groups, 48 were significantly differentially expressed, and among the 28 upregulated in COVID-19 clots were the alpha, beta, gamma, and delta hemoglobin subunits, haptoglobin, biliverdin reductase B, and carbonic anhydrase 1, a signature of free hemoglobin and heme degradation. Because erythrocyte structural proteins such as spectrin, ankyrin-1, and band 3 were not differentially expressed, the hemoglobin within the COVID-19 thrombi appears to have been largely liberated from its cellular packaging, circulating in a cell-free form before being swept into the growing clot. The researchers also found proteins found exclusively in COVID-19 clots: classical complement components including C1QB, C1QC, and C1S, acute-phase reactants such as C-reactive protein and serum amyloid proteins, ubiquitin–proteasome components, and immunoglobulin chains, together painting a picture of intense immune-inflammatory activation and oxidative stress inside the infected thrombi.

How does free hemoglobin promote clotting? The authors draw on a substantial literature showing that cell-free and oxidized hemoglobin are potent procoagulant molecules. Free hemoglobin scavenges nitric oxide, promoting vasoconstriction, and it directly activates platelets by binding glycoprotein Ib alpha, while enhancing von Willebrand factor–mediated platelet adhesion and inhibiting ADAMTS13, the enzyme that normally regulates von Willebrand factor. Heme, its breakdown product, activates endothelial cells through TLR4, stimulates tissue factor expression, and triggers complement deposition on endothelial surfaces. Iron overload further damages endothelial mitochondria through reactive oxygen species. In this framework, the reduced platelet content seen in COVID-19 clots likely reflects platelet consumption and hyperactivation during widespread microthrombus formation rather than a quiet coagulation system, with the surviving platelets showing the morphological stress of filopodia formation observed by electron microscopy.

The source of the free hemoglobin remains an open question. The authors propose that SARS-CoV-2 infection may drive subclinical or compensated hemolysis through cytokine-mediated erythroid injury, complement attack on red blood cells, or direct viral effects on erythrocyte membrane integrity, mechanisms that previous proteomic and lipidomic studies have independently suggested. They caution that their observational design cannot prove causation, that the pre-Omicron viral strains studied may differ from current variants, and that the small sample size and selection of only fully analyzable thrombi impose limitations. Notably, the proteomic signature tracked infection status rather than stroke etiology, suggesting that the hemoglobin-iron enrichment is a distinctive feature of COVID-19 clots independent of the underlying embolic source.

Still, the implications are considerable. Retrieved thrombi are thought to preserve at least part of the molecular milieu present at the moment of clot formation, and the convergence of hemolysis markers, complement activation, antioxidant responses, and platelet depletion in these clots offers a coherent mechanistic account of COVID-19 immunothrombosis that goes beyond the traditional triad of endothelial injury, immune activation, and coagulopathy. If cell-free hemoglobin and iron indeed help drive the prothrombotic state, they become candidate therapeutic targets: haptoglobin infusion, heme-scavenging strategies, or iron-chelating approaches could, in principle, blunt thrombotic risk during severe viral infection. The authors call for larger multicenter studies to directly quantify circulating cell-free hemoglobin and iron-handling biomarkers in COVID-19 patients and to test whether targeting these pathways can reduce ischemic complications, including stroke, and improve the response of these unusually stubborn clots to thrombolysis and mechanical removal.

Subject of Research: Cell-free hemoglobin and iron accumulation in cerebral thrombi as drivers of the SARS-CoV-2-induced prothrombotic state in COVID-19 stroke patients

Article Title: High levels of cell-free hemoglobin and iron in cerebral thrombi of pre-Omicron COVID-19 stroke patients: novel drivers of SARS-CoV-2–induced prothrombotic state

Article References: Payá, M., Alcahut-Rodríguez, C., Barbella-Aponte, R. A., Hernández-Fernández, F., Molina-Nuevo, J. D., Barroso-García, G., Castro-Robles, B., López-López, S., Arias-Salazar, L., Ayo-Martín, Ó., García-García, J., Yélamos-Sanz, B., García-Flores, N., Arandilla, A. G., Moreno-Luna, R., Aliena-Valero, A., Vielba-Gómez, I., Tembl, J. I., Salom, J. B., … Segura, T. (2026). High levels of cell-free hemoglobin and iron in cerebral thrombi of pre-Omicron COVID-19 stroke patients: novel drivers of SARS-CoV-2–induced prothrombotic state. Acta Neuropathologica, 152(1), Article 26. https://doi.org/10.1007/s00401-026-03074-7

Image Credits: AI Generated

DOI: 10.1007/s00401-026-03074-7

Keywords: COVID-19, stroke, cell-free hemoglobin, iron, cerebral thrombi, hypercoagulability, immunothrombosis, proteomics, complement activation, platelets, hemolysis, SARS-CoV-2