Introduction
Cerebral venous thrombosis (CVT) is a rare cause of stroke, with an annual incidence of approximately 3 to 4 per million adults, but is associated with a high mortality rate of 10%-15% and an increasing incidence across a particularly young population [1-3]. One large study cites the incidence of CVT to be as high as 13.9-20.2 per million [4]. The clinical presentation may range from nonspecific headaches and visual obscurations to focal neurologic deficits, seizures, and even coma. Such variability is indicative of the high degree of clinical suspicion necessary for timely diagnosis and management [2,5].
The pathogenesis often reflects the culmination of multiple genetic and acquired risk factors [6]. The classic Virchow’s triad of venous stasis, hypercoagulability, and endothelial changes, establishes the basis for the development of venous thrombi [7]. As such, head trauma, local extension of infections (both central nervous system infections and local parameningeal infections), congenital thrombophilia, and acquired thrombophilia are common etiology. One large national sample study found that a higher mortality rate for CVT is associated with the presence of hematological disorders, systemic malignancy, and central nervous system infections [8]. Another case-control study found malignancy itself to increase the risk of CVT fivefold, especially within the first year of diagnosis and when of hematological etiology [9].
While it has been previously established that malignancy is a risk factor for CVT, specific characteristics regarding comorbidities, severity, and treatment outcomes for CVT patients with cancer remain unclear [10]. Our study aims to elucidate the characteristics, treatments, and outcomes of patients with cancer and CVT (CA-CVT) across the United States.
Methods
Data source and patient selection
The National Inpatient Sample (NIS) database is the largest publicly available national database in the United States and is provided by the Healthcare Cost and Utilization Project [11]. The database encompasses retrospective discharge data from hospitals in 48 states, representing around 35 million hospital discharges across the nation when self-weighted. Approval by institutional review board or patient consent is not required for this study since the NIS contains no identifiable patient information and is a publicly available database. The NIS contains admission and discharge disposition information, and diagnoses and procedures can be queried using the International Classification of Diseases, Tenth Revision (ICD-10) diagnosis or procedure codes.
The NIS 2016-2019 was queried for patients with a diagnosis of CVT as the first listed diagnosis for admission. Patients with cancer were identified using ICD-10 codes that reflect a current diagnosis of cancer. Patients with a diagnosis code suggestive of history of malignancy (ICD-10 Z85) were excluded. CVT patients with cancer (CA-CVT) were compared to CVT patients without cancer (non-CA-CVT).
Data characteristics and outcomes measured
Demographics extracted include age, sex, and race as well as comorbidities of atrial fibrillation, diabetes mellitus, hypertension, hyperlipidemia, chronic heart failure, chronic kidney disease, smoking, anemia, antiplatelet use, long-term anticoagulation, thyroid disease, deep venous thrombosis, liver disease, pulmonary disease, and thrombophilia. Concomitant intracerebral hemorrhage (ICH) upon presentation was defined when ICH was listed as the second or third diagnosis upon admission. Severity was measured using the International Study of Cerebral Vein and Dural Sinus Thrombosis Risk Score (ISCVT-RS) developed by the ISCVT and VENOPORT (Cerebral Venous Thrombosis Portuguese Collaborative Study Group) investigators [12]. This score ranges from 0 to 9, with 0 representing the highest likelihood of a good outcome, and takes into account malignancy, coma, thrombosis of deep venous system, mental status disturbance, male sex, and parenchymal hemorrhage. The score is more likely to predict good outcome; a score of ≤3 depicts the most favorable predictive value and specificity. A detailed description of the score can be found in the original publication [11]. Complications such as hemorrhagic transformation, heparin-induced thrombocytopenia (HIT), percutaneous endoscopic gastrostomy (PEG), tracheostomy, pneumonia, sepsis, and acute kidney injury (AKI) were described. Multivariate regression analysis was used to generate propensity scores controlling for age, sex, insurance status, severity, and the presence of concomitant ICH upon presentation. Using this score, cases were then matched on a 1:1 nearest neighbor system to create cohorts of equal sizes that are controlled for age, sex, insurance status, severity, and ICH. Outcomes of inpatient death, routine discharge, and length of stay were evaluated in this propensity-score matched cohort. Because hematopoietic cancer has a strong association with venous thrombosis, subgroup analyses comparing patients with hematopoietic cancer versus no cancer and non-hematopoietic cancer versus no cancer were evaluated.
Statistical analysis
Categorical variables were compared using Pearson’s chi-squared test; odds ratios and 95% confidence intervals were calculated. Continuous variables were evaluated for normality using the Kolmogorov-Smirnov test. Normally and non-normally distributed continuous variables were tested using the Student’s t-test and Mann-Whitney U test, respectively. Linear regression was also used to evaluate continuous variables. IBM SPSS Statistics software (Version 28.0; IBM Corp., Armonk, NY, USA) was used for analysis, and statistical significance was set at P<0.05.
Results
Characteristics of CVT patients
In the 2016-2019 NIS, 13,955 patients had a diagnosis code of CVT, and 6,140 had a primary diagnosis of CVT. The average age of patients with a primary diagnosis of CVT was 46.0 (±20.3) years old; 4,010 (65.3%) were female; and 2,410 (39.3%) were of nonwhite race. Of the 6,140 patients, 370 (6.0%) had a diagnosis code of malignancy. CA-CVT patients were older than non-CA-CVT patients (55.8±19.8 vs. 45.4±20.1 years; P<0.01) and less likely to be female or of non-white race.
Patients with CA-CVT had higher rates of risk factors for cardiovascular diseases compared to non-CA-CVT patients. CA-CVT patients were more likely to have diabetes mellitus, hypertension, hyperlipidemia, chronic heart failure, and anemia compared to non-CA-CVT patients. There were no differences in rates of atrial fibrillation, chronic kidney disease, smoking, or long-term anti-thrombotic use (Table 1).
Cancer types
Of 370 patients with CA-CVT, 195 (52.7%) of patients had hematopoietic malignancy. This was followed by malignancy of the central nervous system (40 patients, 10.8%), breast (40 patients, 10.8%), and respiratory tract (40 patients, 10.8%) (Table 2).
Hematopoietic cancer
Of all patients with CVT, 195 (3.3%) had hematopoietic cancer (H-CA-CVT). Patients with H-CA-CVT were less likely to be female or on long-term antiplatelet compared to patients without cancer (P<0.01 for both). Patients with H-CA-CVT had higher rates of anemia, deep vein thrombosis, and thrombophilia (P<0.01 for all) (Table 3).
All other cancers, non-hematopoietic
Of all patients with CVT, 175 (47.3%) had cancers other than hematopoietic cancer (non-H-CA-CVT). Patients with non-hematopoietic cancers had higher rates of hypertension, hyperlipidemia, chronic heart failure, deep vein thrombosis, liver disease, and thrombophilia than patients without any cancer (P<0.01 for all) (Table 3).
Severity and complications
Cerebral edema, aphasia, hemiplegia, herniation, and seizure were more likely in CA-CVT patients than non-CA-CVT patients. The average severity score for patients with CVT was 0.64±0.98, and patients with CA-CVT had higher average severity scores than non-CA-CVT patients. There were no differences in rates of coma, mechanical ventilation >24 hours, status epilepticus, or concomitant ICH between the groups (Table 4).
Complications of HIT, PEG, and sepsis were more likely in CA-CVT compared to cancer-free patients. There were no differences in rates of tracheostomy, pneumonia, or AKI between the two groups (Table 4).
Hematopoietic cancer
Compared to patients without malignancy, patients with H-CA-CVT had higher odds of cerebral edema, hemiplegia, coma, mechanical ventilation >24 hours, or concomitant ICH (P<0.05 for all). There were also higher rates of HIT, PEG requirement, sepsis, and AKI in patients with H-CA-CVT versus no cancer (P<0.01 for all) (Table 5).
All other cancers, non-hematopoietic
Patients with non-hematopoietic cancer and CVT had higher rates of hemiplegia (P<0.01) but lower rates of coma (P=0.02) or mechanical ventilation >24 hours (P=0.02). There were higher rates of HIT (P=0.01), PEG (P<0.01), or sepsis (P<0.01) in the patients with non-hematopoietic cancer than those without cancer (Table 5).
Treatment
Thrombectomy was performed in 95 (1.5%) of all patients with CVT, including 5 (1.4%) of CA-CVT patients and 90 (1.6%) of non-CA-CVT patients. There was no statistically significant difference in rates of thrombectomy between the groups. Decompressive hemicraniectomy (DHC) was performed on 25 (0.4%) of CVT patients and 25 (0.4%) of non-CA-CVT patients. No patients with CA-CVT underwent DHC. There was no statistically significant difference between the groups (Table 6).
Outcomes
Patients with CVT and concomitant ICH upon presentation had an increased likelihood of inpatient death and a decreased likelihood of routine discharge. Patients with ICH on presentation also had longer length of stay (8.53±8.68 vs. 5.68±7.64 days, P<0.01) than patients without concomitant ICH. There was no association between concomitant ICH and malignancy (Table 7).
On propensity-score matched analysis, there were no differences in inpatient death, routine discharge, or length of stay between non CA-CVT and CA-CVT patients (Table 7).
Hematopoietic
Patients with H-CA-CVT had higher odds of inpatient death compared to patients without cancer. However, this finding was no longer significant when controlling for severity on propensity-score matching (Table 7).
All other cancers, non-hematopoietic
Patients with non-hematopoietic cancers and CVT were less likely to be discharged home (P=0.02) and had increased likelihood of inpatient death (P<0.01). When controlling for severity, there were no differences between those with non-hematopoietic cancer and no cancer (Table 7).
Discussion
Current studies on CVT in malignancy consist mostly of case series or case-control studies at a few institutions [9,13-15]. There is one large nationwide analysis of the Danish population that shows that patients with CVT were only at a higher risk of a cancer diagnosis within 3 months of CVT [16]. Our study adds to the literature by analyzing the severity, complications, and inpatient outcomes of patients admitted for CVT with or without malignancy.
Apart from the malignancy itself, cancer treatments including steroids and chemotherapeutic agents, and propensity for hospital complications causing infection or venous stasis also increase risk of cerebrovascular events [17,18]. Analyses on a national scale are important to understand the scope of CVT and outcomes in this seldom-studied population [9,19,20].
Patients with CA-CVT had increased severity. As malignancy is a variable in the ISCVT-RS score, it follows that patients in the CA-CVT cohort had much higher scores than those without cancer. However, patients also had higher rates of individual severity characteristics, including cerebral edema, hemiplegia, and seizures. Patients with hematopoietic cancers in particular had higher rates of concomitant ICH. These findings highlight the role that malignancy may play in the development and severity of CVT and can help guide physicians’ understanding of the clinical picture, approach to treatment, and conversations with family.
Increased risk of hospital complications in patients with malignancy may also lead to higher rates of CVT, as sepsis, pneumonia, and requirement of tracheostomy are also known risk factors for venous thrombotic events [21,22]. Patients with CA-CVT had higher rates of hospital complications, such as sepsis, pneumonia, and requirement of tracheostomy, PEG, and AKI, compared to nonCA-CVT patients. In subgroup analyses, patients with hematopoietic cancer and non-hematopoietic cancers also had higher rates of complications. This is consistent with literature showing that the rate of in-hospital adverse events for cancer patients was 24.2% and 17.4% for non-cancer patients and is an important consideration when discussing the association between malignancy and CVT [23].
CVT patients with malignancy were six times more likely than patients without malignancy to experience HIT. Increased rates of HIT were consistent among subgroup analyses of patients with hematopoietic and non-hematopoietic cancers. There is a paucity of literature on HIT in cancer patients, but few reviews and case reports do describe the association between the inflammatory state of malignancy and the development of HIT [24-26]. More focused studies are warranted to assess efficacy of CVT treatments in a population at risk of HIT or bleeds.
We saw an increased mortality rate in patients with malignancy and CVT. We postulate that this is mediated by worse CVT severity rather than a generalized frail state due to malignancy. This is an important distinction when considering treatment options, prognosis, and family discussions. When controlling for severity, there were no differences in outcomes between those with cancer and those without. This is consistent with literature showing similar outcomes in patients with cancer and arterial stroke compared to stroke patients without cancer [27]. However, the NIS is limited to inpatient data, so the outcomes measured—discharge disposition, inpatient death, and length of stay—do not shed light on long-term prognosis or quality of life. In fact, one large study showed that 92% of patients with CVT survived their index admission and elderly patients or patients with a history of cancer had higher rates of 1-year mortality after discharge [28]. Although our paper focuses on inpatient outcomes, the long-term effects of malignancy on patients with CVT are an important area of further study.
As a database analysis, this study has inherent limitations. The NIS captures only inpatient data and lacks measures of functional outcome or degree of disability. We were therefore able to assess neither functional nor long-term outcomes of patients. Additionally, specific data on patients’ pre-hospitalization functional status, cancer severity, or treatment are not available in the database. It is also possible that some patients with a history of cancer or otherwise non-active cancer were miscoded and therefore included in our study, However, we were careful to exclude patients with non-active cancer within the limitations of the NIS. Anticoagulation is the standard of care for CVT. However, initiation of anticoagulation was not available in the database and we therefore could not describe the rate or outcomes of this therapy. Imaging, lab results, or other diagnostic workups are not available in the NIS. We therefore do not know the certain etiologies of CVT in each patient, which is an important factor that affects prognosis. Finally, all diagnoses and procedures are queried using ICD-10 codes, which may be used heterogeneously throughout the nation. Therefore, there may be misclassification or under-representation of CA-CVT patients or procedures such as EVT or DHC.
Conclusions
CVT is an important consideration in patients with malignancy and new-onset neurological symptoms. Patients with hematopoietic cancers, in particular, had increased rates of concomitant ICH. Patients with CVT and cancer may be at risk for more severe presentations and increased mortality. However, when controlling for severity, patients with CA-CVT and non-CA-CVT had similar discharge disposition, inpatient mortality, and length of stay. Further study to elucidate specific cancers and treatment regimens associated with the greatest risk of cerebrovascular events is warranted.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
