Fax: (410) 502-1493
ABO blood group is a potent risk factor for venous thromboembolism in patients with malignant gliomas
Article first published online: 11 MAR 2004
Copyright © 2004 American Cancer Society
Volume 100, Issue 8, pages 1717–1723, 15 April 2004
How to Cite
Streiff, M. B., Segal, J., Grossman, S. A., Kickler, T. S. and Weir, E. G. (2004), ABO blood group is a potent risk factor for venous thromboembolism in patients with malignant gliomas. Cancer, 100: 1717–1723. doi: 10.1002/cncr.20150
- Issue published online: 30 MAR 2004
- Article first published online: 11 MAR 2004
- Manuscript Accepted: 22 JAN 2004
- Manuscript Revised: 12 JAN 2004
- Manuscript Received: 3 NOV 2003
- ABO blood group;
- venous thromboembolism;
- malignant gliomas;
- risk factors
Venous thromboembolism (VTE) is a common cause of morbidity and mortality among patients with malignant gliomas. To investigate the pathogenesis of VTE and facilitate targeted prophylaxis strategies, the authors aimed to characterize VTE risk factors in these patients.
The authors conducted a retrospective chart review of 130 adult patients with glioma who received their primary therapy at the Johns Hopkins Hospital (Baltimore, MD) between 1991 and 2001. Symptomatic VTE was confirmed by objective radiologic testing. The association between clinical and laboratory characteristics and VTE was assessed using parametric and nonparametric statistical tests and survival analysis.
VTE developed in 28 patients (21.5%) at a median of 4.8 months after diagnosis (interquartile range, 2.1–13.2). Patients with tumors > 5 cm were more likely to develop VTE than patients with smaller tumors (hazard ratio = 2.2; P = 0.04). For every year increase in age, the hazard ratios for thrombosis increased by 3% (P = 0.011). When stratified by ABO blood group, the hazard ratios for thrombosis were 2.7 and 9.4 for patients with blood groups A (P = 0.045) and AB (P < 0.0001), respectively, compared with patients with blood group O. No association was observed between VTE and the other patient characteristics analyzed.
Patient age, tumor size, and particularly ABO blood group are risk factors for VTE among patients with malignant gliomas. These findings may facilitate the development of a thrombosis risk score that will allow physicians to individualize VTE prophylaxis regimens. Cancer 2004. © 2004 American Cancer Society.
Venous thromboembolism (VTE) is a common complication of cancer.1 Contemporary estimates indicate that 15% of patients with cancer develop a VTE during the course of their disease and 50% have evidence of thrombosis after a postmortem examination.2, 3 Patients with malignant gliomas are at particularly high risk for developing VTE. Previous reports indicated that 19–28% of patients with glioma develop symptomatic venous thrombosis during their clinical course.4–8 Despite the high prevalence of thrombotic disease, few consistent risk factors for VTE have been identified among patients with glioma. Previous studies have identified patient age,4, 9 tumor size,10 tumor grade,8 prolonged operative times (> 4 hours),11 chemotherapy,6 and the presence of leg paresis4, 6, 8 as possible clinical risk factors for the development of VTE. Laboratory variables associated with VTE include increased prothrombin time (PT) and reduced plasminogen activator activity.12, 13 Only age and leg paresis were risk factors for VTE in more than one study. Because patients with brain tumors are believed to be at high risk for hemorrhagic complications of anticoagulation,14 the identification of reproducible risk factors for VTE may allow more discriminative use of anticoagulant approaches to VTE prophylaxis. In the current study, we report the clinical and laboratory parameters that influence the development of thrombosis in a retrospective cohort study of patients with glioma who received their primary therapy at the Johns Hopkins Hospital (Baltimore, MD).
MATERIALS AND METHODS
As part of a retrospective immunohistochemical study of tissue factor expression in patients with malignant gliomas,15 a search of the pathology data system at The Johns Hopkins Hospital identified adult patients (≥ 18 years of age) who had been newly diagnosed with a malignant glioma between 1991 and 2001. A subset of these patients, whose pathologic material was available for review, was selected for a retrospective analysis of medical records. For each patient, demographic characteristics as well as a predetermined set of clinical and laboratory variables potentially associated with the development of VTE were abstracted and entered into an Excel database (Microsoft, Redmond, WA). Clinical variables included age, gender, ethnicity, body mass index (BMI), tumor size (measured by a gadolinium-enhanced magnetic resonance imaging scan) and grade (scale 1 to 4), the presence of extremity paresis (defined as limb strength of ≤ 3 on a 0–5 scale during physical examination), operative time (defined as the time from induction of anesthesia to completion of the case; in the case of multiple surgeries, the longest recorded operative time was used), exposure to systemic or localized chemotherapy, and overall and thrombosis-free survival. Laboratory variables included ABO blood type, baseline hemoglobin level, platelet count, PT, and activated partial thromboplastin time (aPTT).
The diagnosis of deep venous thrombosis (DVT) required objective radiologic documentation using duplex ultrasonography or venography. The diagnosis of pulmonary embolism required documentation of a high-probability ventilation perfusion scan, a positive spiral computerized tomography scan, or pulmonary angiography. Corticosteroids and antiepileptic medications were administered as dictated by the clinical situation at the discretion of the patient's physician. The Johns Hopkins University internal review board approved all methods of data acquisition and analysis.
The primary outcome variable in the study was the development of VTE. Mean or median values of baseline clinical and laboratory characteristics were compared among patients with and without thrombosis using the Student t test or Wilcoxon rank-sum tests, as appropriate. Patients were categorized into quartiles for age, operative time, tumor size, BMI, baseline platelet count, baseline hemoglobin level, and platelet count at the last measurement before thrombosis. Additional categories were created for baseline international normalized, ratio and baseline aPTT to capture the lowest 25%, middle 50%, and top 25% of these distributions.
Patients were censored at the time of death or the time at which the patient was lost to clinical follow-up, if thrombosis had not occurred. Kaplan–Meier curves were plotted with the patients stratified by these categorical variables and differences in the duration of thrombosis-free survival were tested with the log-rank test. Cox regression analysis was used to test the influence of continuous variables on the time to thrombosis. In addition, we used Cox regression to evaluate variables that were time varying and included a term for clustering by patient.
For variables that significantly influenced time to thrombosis, as demonstrated with the log-rank test or the Cox models, further analysis was performed using Weibull and Cox models after testing for proportionality of the hazards and model fit. Outcomes from these models are reported as the relative hazards of thrombosis for one group compared with another. All statistical analysis was performed using Stata software (version 7.0; College Station, TX).
One hundred thirty individuals constituted the study population. The median age at diagnosis was 55 years (Table 1). Males comprised 64% of the study population and whites comprised 87%. The initial histologic diagnosis was glioblastoma multiforme for 83% of patients. The median tumor size was 5 cm (range, 2–9 cm.). Perioperative VTE prophylaxis consisted of sequential compression devices and thromboembolism (TED) stockings for nearly all patients. TED stockings alone were used for only one patient. A few patients (21%) received additional prophylaxis in the form of low-dose subcutaneous unfractionated heparin (5000 IU twice per day). After macroscopic total resection, all patients received adjuvant radiotherapy. Forty-eight percent received systemic chemotherapy and 31% were treated with local chemotherapy (polifeprosan 20 with carmustine [Gliadel]; Guildford Pharmaceutical, Baltimore, MD) during their clinical course. Baseline laboratory values including median hemoglobin level, platelet count, PT, and aPTT were within the normal range for all patients (Table 2). The median survival of the cohort was 13 months (interquartile range 8–21 months) and the median duration of clinical follow-up was 12 months (range, 6.5–21 months).
|Patient characteristics||Value (IQR)|
|Median age at diagnosis (yrs)||55 (47–66)|
|Ethnicity (% of cohort)|
|Median BMI at diagnosis (kg/m2)||25.8 (24–28.3)|
|Tumor size (cm)||5 (4–6)|
|Tumor grade, scale 1–4 (%)|
|Grade 2 glioma||1.5|
|Grade 3 (anaplastic astrocytoma)||15.5|
|Grade 4 (glioblastoma multiforme)||83|
|Median operative time (min)||300 (255–335)|
|Local (polifeprosan 20 with carmustine)||31|
|Laboratory test||Value (IQR)|
|Hemoglobin (g/dL)||14 (12.2–14.9)|
|Platelet count (× 103/μL)||256 (202–298)|
|PT (sec)||12.1 (11.5–12.6)|
|aPTT (sec)||23 (21.2–25.8)|
|aPTT ratio||0.9 (0.8–0.9)|
VTE developed in 28 of 130 (22%) patients during their clinical course. The median time to thrombosis was 4.8 months (range, 2.1–13.2 months) after diagnosis. The results of univariate analysis are shown on Tables 3 and 4. Three characteristics were associated with an increased risk of venous thrombosis. When stratified by age into quartiles, patients age < 45 years had a significantly longer thrombosis-free survival compared with older subjects (P = 0.026) (Fig. 1). When age was modeled as a continuous variable, the hazard of VTE increased by 3% for every 1-year increase in age (P = 0.011). Patients with tumors > 5 cm in greatest dimension also had significantly reduced thrombosis-free survival (P = 0.038) (Fig. 2). When patients were stratified by tumor size, the hazard ratio for thrombosis was 2.2 (95% confidence interval ([95% CI], 1.0–4.5)) for patients with tumors > 5 cm. compared with patients with smaller tumors (P = 0.043).
|Characteristics||Patients without VTE (n = 102) (IQR)||Patients with VTE (n = 28) (IQR)||P value|
|Median age (yrs)||56 (44–66)||54 (51–62)||1.0|
|Median survival (mos)||12 (8–24)||13 (9–20)||0.4|
|Male gender (%)||63||68||0.6|
|Tumor size (cm)||5 (4–6)||5.1 (4–6)|
|BMI (kg/m2)||26 (24–28)||26 (25–29)||0.7|
|Limb paresis (%)||41||50||0.4|
|Local (polifeprosan 20 with carmustine)||28||36||0.5|
|Laboratory test||Patients without VTE (n = 102) (IQR)||Patients with VTE (n = 28) (IQR)|
|Hemoglobin (g/dL)||14 (13–14.9)||13.9 (13.2–14.6)|
|Platelet count (× 103/μL)||254 (201–302)||262 (213–290)|
|PT (sec)||12.1 (11.5–12.6)||12.1 (11.4–12.8)|
|INR||1 (1–1)||1 (0.9–1)|
|APTT (sec)||23.2 (21.2–25.9)||22.3 (20.9–25.1)|
|APTT ratio||0.9 (0.8–0.9)||0.9 (0.8–0.9)|
Striking differences in thrombosis-free survival were observed when the cohort was stratified by ABO blood group (Fig. 3). The hazard ratio for thrombosis among patients with blood group AB was 9.4 (95% CI, 2.7–32) compared with patients with blood group O (P < 0.0001) and the hazard ratio among patients with blood group A was 2.7 (95% CI, 1.0–7.0]) compared with those with blood group O (P = 0.045). The category of patients with blood group B violated the proportional hazards assumption such that a hazard ratio could not be reported for this patient group. No other clinical or laboratory characteristics including the presence of extremity paresis or administration of chemotherapy were associated with the development of thrombotic complications.
VTE is a well documented source of morbidity and mortality among patients with malignant gliomas.4–8, 16 Despite the routine use of mechanical VTE prophylaxis in the perioperative period, 22% of the patients in our cohort study suffered a thrombotic event during the course of their illness. These findings are similar to those observed in previous retrospective studies, which have identified clinical thrombotic events in as many as 28% of patients with glioma.4–7 Moreover, 125I-labeled fibrinogen leg scans have been shown to identify symptomatic and/or subclinical venous thromboses in 60% of patients with glioblastoma during the postoperative period.10 In what to our knowledge is the only published prospective study of patients with glioma, a 32% risk of DVT was observed 24 months after surgery. Although prophylactic anticoagulation has proven efficacy in the prevention of venous thrombosis in neurosurgical patients,17–19 widespread use remains limited due to concerns regarding bleeding complications.14 These reports highlight the impact of hypercoagulability in neurosurgical patients and emphasize the need to develop clinical strategies to prevent thromboses. The identification of VTE risk factors, which could be incorporated into a validated VTE risk profile, would greatly enhance efforts to identify patients with glioma who are at the greatest risk for VTE and in whom the risk-benefit ratio would favor more aggressive prophylactic measures.
The results of the current study suggest that ABO blood group status is a potent and previously unrecognized risk factor for the development of thrombosis in patients with malignant glioma. Evidence suggests that this finding is compatible with modern concepts of hemostasis. ABO blood type has a profound influence on von Willebrand factor (vWF) and factor VIII activity levels. The lowest serum levels of these proteins have been associated with blood group O and the highest levels with blood group AB, whereas intermediate serum levels have been associated with blood groups A and B.20, 21 Elevated levels of factor VIII, which is stabilized in the circulation by vWF, have been shown to increase the risk of both initial and recurrent VTE by as much as four to six fold.22, 23 Although the direct association between serum factor VIII activity and the risk of thrombosis in patients with glioma remains to be determined, our results suggest that ABO blood group status is a potential marker for this association. Unlike previously identified markers of VTE risk, ABO blood type is routinely determined preoperatively by objective and standardized methods and would be readily available to guide early decision-making regarding VTE prophylaxis.
Similar to previous studies of VTE in patients with glioma,4, 9, 10 we also identified patient age and tumor size as risk factors for VTE. Serum levels of procoagulant factors generally increase with age whereas anticoagulant protein levels remain stable or slightly decrease, which may contribute to the higher prevalence of VTE among older individuals and explain the trend observed by us and others.24–26 Although to our knowledge the etiology of the procoagulant state in patients with glioma remains incompletely understood, tumor products such as tissue factor and plasminogen inhibitors likely play an important role in the pathogenesis of VTE associated with gliomas.13, 27, 28 Because larger tumors may be associated with greater production of these substances, a correlation with tumor burden is biologically plausible.
In contrast, we were not able to confirm the reported associations between the development of VTE and extremity paresis, tumor grade, exposure to chemotherapy, prolonged operative times, and a prolonged PT.4, 6, 8, 11, 12 Unlike previous investigations, we used a strict definition of paresis based on the reported results of physical examinations. Although we believe this design made our approach to paresis more uniform, it may be responsible for our dissimilar results compared with previous investigations. Although unproven, differences in subject populations, VTE prophylactic and glioma therapeutic regimens, and outcomes assessment are readily apparent between the current study and previous studies. This may account for our inability to identify an influence of tumor grade, chemotherapy, operative time, or PT on VTE risk. These inconsistencies underscore the need for a large prospective study to validate previously identified VTE risk factors.
Although the postoperative period clearly is associated with an increased risk of thrombotic events, similar to previous studies,5, 8 the current study finding that the median time to thrombosis was 4.8 months indicates that the risk of VTE persists throughout the clinical course of patients with malignant gliomas. Therefore, prophylactic efforts focused on the immediate postoperative period may provide insufficient protection for patients with malignant gliomas.
Because the current study was conducted retrospectively on a segment of the glioma population treated at the study institution over a 10-year period, its inherent limitation is the lack of prospective validation. It is conceivable that the prevalence of VTE in the current study population was underestimated because routine radiologic surveillance for VTE was not performed and investigation for vascular events was performed at the discretion of individual physicians. Nonetheless, the results of the current study legitimately reflect clinical practice and should serve as a foundation for future confirmatory studies. The patient population we investigated was selected from the available cases in the pathology data system by two of the investigators (M.B.S. and E.G.W.) who had no previous knowledge of clinical course, thereby minimizing the likelihood of selection bias. The primary outcome measure was determined by using objective radiologic techniques, which minimized the likelihood for misclassification of thrombotic events. Furthermore, strict definitions were applied to all exposure variables to ensure uniformity of classification. We believe these findings are biologically tenable and are supported by existing data in the literature.
VTE is a common event in patients with malignant gliomas that occurs throughout the course of the disease. Although we have shown that patient age and tumor size are significantly associated with the development of thrombosis in patients with glioma, the data from the current study indicate that ABO blood group status is a particularly potent risk factor for VTE. Prospective studies investigating the utility of these risk factors for identifying patients with glioma at high risk for thrombosis are warranted and should allow for the development of a VTE risk score that will permit physicians to individualize VTE prophylaxis regimens. A similar VTE risk score may impact the clinical management of patients with other malignancies known to be associated with a high incidence of thrombotic complications. Our results suggest that ABO blood group antigens may play a role in the pathogenesis of thrombosis in patients with gliomas and other malignancies, possibly through interactions with known procoagulant proteins.
- 3Abnormalities of hemostasis in malignant disease. In: ColmanR, HirshJ, MarderV, SalzmanE, editors. Hemostasis and thrombosis. Philadelphia: J.B. Lippincott, 1994: 1238–1254..
- 15Correlation between tumoral tissue factor expression and the incidence of venous thromboembolism in patients with malignant gliomas [abstract]. Blood. 2002; 100: 1349a., , , et al.