J Neurooncol (2015) 124:87–94 DOI 10.1007/s11060-015-1805-2
CLINICAL STUDY
Venous thromboembolism (VTE) and glioblastoma Shlomit Yust-Katz1,2 • Jacob J. Mandel3 • Jimin Wu5 • Ying Yuan5 • Courtney Webre4 • Tushar A. Pawar6 • Harshad S. Lhadha6 • Mark R. Gilbert6 Terri S. Armstrong6,7
•
Received: 29 October 2014 / Accepted: 6 May 2015 / Published online: 19 May 2015 Ó Springer Science+Business Media New York 2015
Abstract The risk of venous thromboembolism (VTE) is high for patients with brain tumors (11–20 %). Glioblastoma (GBM) patients, in particular, have the highest risk of VTE (24–30 %). The Khorana scale is the most commonly used clinical scale to evaluate the risk of VTE in cancer patients but its efficacy in patients with GBM remains unclear. The aim of this study is to estimate the frequency of VTE in GBM patients and identify potential risk factors for the development of VTE during adjuvant chemotherapy. Furthermore, we intend to
Shlomit Yust-Katz and Jacob J. Mandel author have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s11060-015-1805-2) contains supplementary material, which is available to authorized users. & Shlomit Yust-Katz [email protected] 1
Neuro-Oncology Unit, Davidoff Cancer Center, Rabin Medical Center, 39 Jabotinski St., 49100 Petah Tikva, Israel
2
Sackler Faculty of Medicine, Tel Aviv University, Ramat Aviv, Tel Aviv, Israel
3
Department of Neurology, Baylor College of Medicine, 6550 Fannin Suite 1801, Houston, TX 77030, USA
4
College of Medicine, Texas A&M Health Science Center, 8447 Highway 47, Bryan, TX 778071, USA
5
Department of Biostatistics College of Medicine, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 1411, Houston, TX 77030, USA
6
Department of Neuro-Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 431, Houston, TX 77030, USA
7
Department of Family Health, The University of Texas Health Science Center-Houston SON, 6901 Bertner Ave, Houston, TX 77030, USA
examine whether the Khorana scale accurately predicts the risk of VTE in GBM patients. We retrospectively reviewed the medical records of GBM patients treated at MD Anderson during the years 2005–2011. The study cohort included 440 patients of which 64 (14.5 %) developed VTE after the start of adjuvant treatment. The median time to develop VTE was 6.5 months from the start of adjuvant treatment. On multivariate analysis male sex, BMI C 35, KPS B 80, history of VTE and steroid therapy were significantly associated with the development of VTE. The Khorana scale was found to be an invalid VTE predictive model in GBM patients due to poor specificity. Of the 64 patients who developed a VTE, 36 were treated with anticoagulation, 2 with an IVC filter, and 21 with both. Complications (intracranial hemorrhage, bleeding in other organs and thrombocytopenia) secondary to anticoagulation were reported in 16 % (n = 10). VTE is common in patients with GBM. Our results did not validate the Khorana scale in GBM patients. Additional studies identifying which GBM patients are at highest risk for VTE are needed to enable further evaluation of VTE preventive measures in this selected group. Keywords Deep vein thrombosis Pulmonary emboli Brain tumors
Introduction The risk of venous thromboembolism (VTE) is increased in patients with solid tumor malignancies [1]. Different types of solid cancers have shown varying risks for the development of VTE. Brain tumors, including glioblastoma (GBM), are one of the highest risk cancers for the development of VTE. A VTE prevalence of approximately 24–30 % (over the full trajectory of the illness) has been reported in high
123
88
J Neurooncol (2015) 124:87–94
grade glioma patients [2–4]. Several risk factors have previously been associated with the development of VTE in GBM patients including functional status, paresis, residual disease, obesity, chemotherapy and vascular endothelial growth factor targeted treatment [5]. It is also well established that the risk for VTE is high in the post surgical period and can extend up to 30 days after the surgical procedure [6]. However, it is unclear if an increased risk of VTE remains after the post-surgical period. Additionally, it is unknown if the risks associated with developing a VTE after surgery are similar to the risks of developing a VTE during adjuvant treatment with chemotherapy. Determining which patients are at a higher risk of VTE is essential as VTE occurrence can cause significant morbidity or mortality and delay vital chemotherapy. Furthermore, distinguishing patients at high risk for VTE could help improve the design of future clinical studies examining the efficacy of VTE prophylactic measures (including anticoagulation) and potentially prevent the hazard of anticoagulation in patients who do not require it [7, 8]. The most commonly used clinical scale to evaluate the risk of VTE development among cancer patients is the Khorana Chemotherapy Associated VTE Predictive Scale. The Khorana scale was developed for patients with a wide spectrum of primary cancers and includes the patient’s type of cancer, platelet count, hemoglobin level, use of erythropoietin stimulating agents, leukocyte count and body mass index(BMI) [9] (Table 1). This study estimates the frequency and examines the potential risk factors of patients with GBM who developed VTE after the start of adjuvant chemotherapy and were followed and/or treated at the MD Anderson Cancer Center. Additionally, we tested whether the Khorana scale accurately predicts the risk of VTE in GBM patients.
Materials and methods Patients After securing institutional review board approval (protocol PA12-0894), we conducted a retrospective chart review Table 1 The modified Khorana scale
of patients with primary GBM treated at MD Anderson from 2005 to 2011 and identified patients who developed a VTE during the course of their disease. All patients had a pathologically-confirmed diagnosis of primary GBM using WHO criteria. Diagnosis of VTE was determined by either ultrasound examination of the extremities, radiographic findings on CT (computed tomography) angiography of the chest or MR (magnetic resonance) venous angiography of the brain. All VTE cases identified were symptomatic (leg swelling, hand swelling, headache, dyspnea, or tachycardia). For our risk factor analysis of developing a VTE after the start of adjuvant chemotherapy, we excluded patients if they did not receive treatment with the established standard of care following surgery of concurrent chemoradiation with daily temozolomide for 6 weeks, developed a VTE before the start of adjuvant treatment or were not followed for at least 6 months after the start of adjuvant treatment. Data for the risk factor analysis (demographic and clinical characteristics) was collected on all eligible patients from the visit prior to starting the first cycle of adjuvant chemotherapy. Data regarding treatment for GBM was collected from follow-up clinic visits. Survival data was also collected from the patient’s clinical charts (date of death is documented in our institutional records). Furthermore, patients who developed a VTE after starting adjuvant treatment also had additional data collected at the time of VTE diagnosis. The additional data at time of VTE included date of VTE, type of VTE (deep vein thrombosis, pulmonary emboli, or sinus venous thrombosis), location of VTE, performance status, steroid treatment, tumor status (stable, progressing), weight, chemotherapy at time of VTE, treatment for VTE and complications of treatment. Statistical analysis Data was first summarized using standard descriptive statistics for demographic and clinical variables. Survival and time from GBM diagnosis to VTE diagnosis were evaluated using the Kaplan–Meier method and the comparison between or among patients’ characteristics was assessed using log-rank test [10]. Both univariate and multivariate Cox regression models were applied to assess
Patient characteristics
Risk score
Site of cancer Very high risk (brain, stomach, pancreas) High risk (lung, lymphoma, gynecologic, bladder, testicular)
1
Pre-chemotherapy platelet count C350,000/mm3
1
Hemoglobin level \10 g/dl or use of red cell growth factors
1
Pre-chemotherapy leukocyte count [11,000/mm3
1
2
Body mass index C35 kg/m
High-risk score C 3; Intermediate-risk score = 1–2; Low-risk score = 0
123
2
1
J Neurooncol (2015) 124:87–94
the effect of covariates of interest on the risk to develop VTE. All statistical analyses were carried out in SAS 9.3 (SAS Institute Inc., Cary, NC, USA). The Khorana Chemotherapy Associated VTE Predictive Scale predicts the risk of developing VTE in cancer patients [7–9]. It is assessed according to several risk factors (see Table 1). We applied the modified Khorana scale that includes high, intermediate, and low risk groups to the collected data in order to classify patients into those groups and then calculated its sensitivity and specificity. Khorana et al. reported that the Khorana scale has a sensitivity of 40.0 % and a specificity of 88 % based on a validation cohort of 1365 patients [9]. If the sensitivity and specificity of the Khorana scale in our collected data is comparable to these reported values then we will regard the Khorana scale applicable to the GBM patient population.
Results A total of 543 patients with primary GBM treated at MD Anderson during the years 2005–2011 were identified. Of these patients, 117 (22 %) were diagnosed with VTE during the course of their illness. 64 (55 %) patients developed VTE after starting adjuvant chemotherapy and 53 (45 %) developed VTE before starting adjuvant treatment [20 (17 %) during the post operative period and 33 (28 %) during radiation or during the time between radiation and the start of adjuvant treatment]. The target population for the risk factor analysis of our study was patients who developed VTE after starting adjuvant chemotherapy. Therefore, in our analysis, we excluded patients that developed VTE before starting adjuvant treatment (53 patients), as well as patients who did not receive the current standard of care with concurrent radiation and temozolomide (TMZ) prior to starting adjuvant chemotherapy treatment or were not followed for at least 6 months after starting adjuvant treatment (50 patients). Thus, the study sample included 440 patients and 64 (14.5 %) of them developed a VTE. The clinical characteristics of our study population are summarized in Table 2 and in the supplementary data (Table 1s). Patient’s characteristics at time of VTE diagnosis Sixty four patients developed VTE after initiation of adjuvant chemotherapy. Their characteristics are described in in the supplementary data (Tables 2s and 3s). 15 patients developed pulmonary emboli (PE), 27 had deep vein thrombosis (DVT), 20 developed both a DVT and PE, and sinus vein thrombosis was diagnosed in 2 patients. Of the 47 patients who developed a DVT, two (4 %) developed a DVT in the upper extremities. Both of these
89
patients had central lines and the DVTs occurred on the same side as the central line. Median age at time of VTE was 56 (range 19–80). The median time to develop VTE was 6.5 months (Range: 0.03–120.74). Thirty (46.9 %) patients developed VTE within 6 months after starting adjuvant chemotherapy. 60 % (37) of patients were ambulatory without an assistive device at time of VTE diagnosis and 30 % (19) of the patients had evidence of hemiparesis on exam. The median karnofsky performance score (KPS) at time of VTE diagnosis was 80 (20–90). The majority of patients with VTE [72.6 % (45)] were on steroids at time of VTE diagnosis. Eighty one percent (52) of patients were being treated with chemotherapy at the time of VTE diagnosis. 48 % (25) were on a TMZ containing regimen, 23 % (12) on a bevacizumab containing regimen, 4 % (2) on both TMZ and bevacizumab, and 25 % (13) on various other chemotherapies. There was evidence of clinical or radiographic progression in 48 % (31) of patients at time of VTE diagnosis. Three patients (4.7 %) were classified later as having had pseudoprogression. Of the 64 patients who developed a VTE, 36 were treated with anticoagulation alone (8 patients with Coumadin and the rest with low molecular weight heparin), 2 had an inferior vena-cava (IVC) filter alone, and 21 patients received both an IVC filter and anticoagulation. Treatment was unknown for 3 patients and another 2 patients were not treated as they were referred to hospice. 40 % of patients with a VTE were admitted for treatment. Most of the patients (93 %) receiving anticoagulation were treated for greater than 6 months. The decision to insert an IVC filter at time of VTE diagnosis was determined by the treating physicians preference based upon history of a hemorrhagic event prior to VTE diagnosis, development of a complication while on anti-coagulation, or desire to delay anticoagulation due to a recent or upcoming surgical resection. Complications secondary to anticoagulation were reported in 16 % (n = 10) of patients. These complications included brain hemorrhage in 3 patients (subdural and intracerebral), bleeding to other organs in 3 patients (rectal sheath, rectal and bleeding to Port- A-Cath), and thrombocytopenia in 4 patients (one who was diagnosed with heparin induced thrombocytopenia). Four of the patients that developed complications secondary to treatment (two with intracranial bleeding, one with a rectal sheath hematoma and one with heparin induced thrombocytopenia) were sent to hospice after admission for the complication. Only 2.5 % (n = 11) of patients had a history of VTE before diagnosis. A very small minority of the patients had a history of stroke (1.1 % n = 5), myocardial infarct (2 %
123
90
J Neurooncol (2015) 124:87–94
Table 2 Summary of demographic and clinical characteristic of patients diagnosed with venous thromboembolism at time of venous thromboembolism diagnosis -categorical covariate Covariate
Levels
Patient diagnosed with VTE Number of events Location of VTE
Pulmonary embolus (PE) Number of brain surgeries before VTE diagnosis
Number of hospitalizations in the last 6 months
Steroid therapy
Tumor status at the time of thrombosis diagnosis
KPS
Paresis
Ambulatory
BMI
Chemotherapy at time of VTE
123
N (%) 64 (100 %)
1
63 (98 %)
2
1 (1.6 %)
One leg
38 (59 %)
Two legs
7 (11 %)
Hand
2 (3.2 %)
PE with no proven evidence of DVT Sinus vein thrombosis
15 (23.8 %) 2 (3.2 %)
No
29 (45.3 %)
Yes
35 (54.7 %)
1
37 (57.8 %)
2
20 (31.3 %)
3 and more
7 (11 %)
Unknown
2
0
16 (25.8 %)
1
34 (54.8 %)
2 or more
12 (19 %)
Unknown
3
No
16 (25.8 %)
Yes
45 (73.6 %)
Stable
26 (40.6 %)
Radiographic progression only Radiographic progression and clinical progression
11 (17.2 %) 23 (35.9 %)
Clinical progression
4 (6.3 %)
Unknown
9
B80
42 (80 %)
90
13 (20 %)
Unknown
2
None
33 (53.2 %)
Upper limb paresis or plegia
3 (4.8 %)
Lower Limb Paresis
2 (3.2 %)
Hemiparesis
19 (30.6 %)
Paraplegia
1 (1.6 %)
Hemiplegia
3 (4.8 %)
Quadriplegia
1 (1.6 %)
Unknown
3
No
13 (21.3 %)
Ambulatory without aid Ambulatory with walker
37 (60.7 %) 11 (18 %)
Unknown
7
\35
45 (78.9 %)
C35
12 (21.1 %)
Unknown
2
None
10 (16.1 %)
Bev alone
1(1.6 %)
Bev plus other chemotherapy
11 (17.7 %)
Tmz alone
10 (16.1 %)
Tmz plus other chemotherapy
11 (17.7 %)
J Neurooncol (2015) 124:87–94
91
Table 2 continued Covariate
Central venous catheter
Treatment for VTE
Duration of treatment
Complications of treatment
Admitted for VTE
Levels
N (%)
Tmz plus thalidomide regimen
4 (6.5 %)
Tmz plus bev
2 (3.2 %)
Other
13 (21 %)
Unknown
2
No
57 (91.9 %)
Yes
5 (8.1 %)
Unknown
3
Anticoagulation only
36 (59 %)
IVC filter only
2 (3 %)
Anticoagulation and IVC filter
21 (34 %)
Hospice and no treatment Unknown
2 (3 %) 21
6 Months or more
40 (93 %)
1 Month or less
3 (7 %)
No complications
54 (84.3 %)
Thrombocytopenia
4 (6.3 %)
Brain hemorrhage
3 (4.7 %)
Bleeding other
3(4.7 %)
Unknown
22
No
25 (59.5 %)
Yes
17 (40.5 %)
VTE venous thromboembolism, PE pulmonary emboli, BMI body mass index, KPS karnofsky performance status
n = 9) or atrial fibrillation (1.4 %, n = 6). History of smoking was positive in 8.9 % (n = 39) of the patients. 90 % (n = 393) of patients were able to ambulate without assistance at the time of starting adjuvant treatment. Only 34 % (145) of patients were on steroids at the time of starting adjuvant treatment. A complete description of the concomitant medication that patients were on at the time of starting adjuvant treatment is listed in the supplementary data (Table 1s). A variety of dosing regimens were used for adjuvant TMZ. Most patients (60 %, n = 243) were treated with the Stupp regimen (5/28). At the start of adjuvant chemotherapy, a high percentage of patients (53.2 %, n = 218) were treated with other chemotherapeutic agents in addition to adjuvant TMZ usually as part of clinical trials. 67 % (n = 295) of patients were treated with bevacizumab during the course of their illness. A small percentage (6.8 %, n = 30) of patients were not treated with adjuvant TMZ due to either tumor progression after chemoradiation or low blood counts. Some of these patients were treated with other adjuvant treatments after chemoradiation. Several patients (3.9 %, n = 17) were being treated with anticoagulation at the time of GBM diagnosis for a variety of reasons including history of VTE and atrial fibrillation. An additional 25 patients (5.7 %) were started on 1 mg of coumadin prophylactically at the time of
adjuvant therapy as part of a clinical trial containing thalidomide. Risk factors for the development of VTE We next assessed which factors were associated with risk to develop VTE by univariate analysis. The significant categorical covariates were sex (p = 0.0102), history of venous thrombosis (p = 0.0056), KPS at start of adjuvant TMZ (p = 0.0316), paresis (p = 0.0004), ambulatory status (p = 0.0138), and BMI (p = 0.0047). Concomitant medications that were associated with risk of VTE included anticoagulation (p = 0.0105), anti-hypertensives (p = 0.0264), and steroid therapy (p \ 0.0001). Anti- gastroesophageal reflux drugs (p = 0.0106) correlated with the risk of VTE but were also found to be associated with corticosteroid use. The significant continuous covariates were age at diagnosis (p = 0.0448), KPS at start of adjuvant TMZ (p = 0.0028) (KPS was assessed as continuous and then categorized), pre-adjuvant TMZ white blood cell count (WBC) (p = 0.0010), pre- adjuvant TMZ absolute neutrophil count (ANC) (p = 0.0001), and steroid dose (p = 0.0014). Factors that were significant (p value \ 0.05) in the univariate cox regression analysis were then taken into a
123
92
J Neurooncol (2015) 124:87–94
multivariate cox regression model. On multivariate analysis, sex (p \ 0.0122), BMI (p \ 0.0145), KPS at start of adjuvant TMZ (p \ 0.0125), history of VTE (p \ 0.0031) and steroid therapy (p \ 0.0115) remained significant (Table 3). Khorana scale We then evaluated whether the Khorana scale is predictive for thrombosis in the GBM population. Using the Khorana scale, all GBM patients were assigned to the very high risk category (risk score = 2) in terms of the cancer site and therefore required only 1 additional risk factor to reach the total risk score C3 to be considered as high-risk overall. As a consequence, most of the patients (n = 418) were categorized as high risk when the scale was applied (Table 4). 63/64 patients that developed VTE were in the high risk group for a sensitivity of 98 %. However, only 21/376 patients that did not develop VTE were in the intermediate risk group for a specificity of 5.6 %.
Discussion This study confirms the high prevalence of VTE in GBM patients with 22 % of GBM patients diagnosed with a VTE. In addition, we found that VTEs commonly occur during adjuvant chemotherapy (64, 55 %) with similar incidence to the postoperative period (53, 45 %). Our study confirms that obesity, history of VTE, elevated WBCs and low ANCs correlate with a higher risk of VTE in GBM
patients. Additionally, several factors that are more specific for GBM patients including functional status and steroid treatment were also found to correlate with VTE risk. The Khorana scale was found to be lacking adequate specificity to predict the risk of VTE in this patient population and therefore is not an adequate tool to predict the risk of VTE in GBM patients. The Khorana scale may not be valid in GBM patients for several reasons. Perhaps, due to the high risk associated with the GBM diagnosis alone, the importance of other specific risk factors are not accurately differentiated or weighed. Another possibility is that the factors included in the Khorana scale may not be relevant in the GBM population. Distinct from other cancers, brain tumor patients often have a poor performance status and increased steroid requirements, both of which have been associated with increased VTE risk [5]. Notably, the dataset from which the Khorana scale was developed included only a few brain tumor patients (n = 4) and patients with poor performance status were underrepresented [9]. This study is one of the first to evaluate the Khorana scale utility in this high risk patient population. A variety of clinical risk factors including age, obesity, medical comorbidity, prior episodes of thrombosis, and hematological parameters, have been associated with development of VTE among cancer patients [8]. Our study confirmed that obesity, history of VTE, elevated WBCs, and low ANCs correlate with a higher risk of VTE in GBM patients. Additionally, several factors that are more specific for GBM patients, including functional status and steroid treatment were also found to correlate with VTE risk. Notably, 72.6 % of GBM patients who developed VTE
Table 3 Multivariate Cox regression model Parameter
Level
Hazard ratio
95 % Confidence limits
p value
Sex
Male versus female
2.33
1.20
4.50
0.0122
History of VTE
Yes versus No
4.50
1.66
12.19
0.0031
KPS at start of adjuvant TMZ
50–80 versus 90–100
2.22
1.19
4.15
0.0125
Paresis BMI
Others versus none C35 versus \35
1.18 2.65
0.54 1.21
2.58 5.80
0.6841 0.0145
Ambulatory status
Others versus none
0.73
0.09
6.02
0.7705
Blood pressure medication
Yes versus No
1.06
0.56
2.03
0.8564
GERD medication
Yes versus No
1.08
0.59
1.98
0.7927
Steroid therapy
Yes versus No
2.20
1.19
4.05
0.0115
Age at diagnosis
1.01
0.98
1.03
0.6845
WBC at start of adjuvant TMZ
0.91
0.67
1.23
0.5385
ANC at start of adjuvant TMZ
1.24
0.90
1.71
0.1942
VTE venous thromboembolism, KPS karnofsky performance status, BMI body mass index, GERD gastroesophageal reflux disease, WBC white blood cell count, ANC absolute neutrophil count
123
J Neurooncol (2015) 124:87–94
93
Table 4 Applying the Khorana scale for the study population VTE
Khorana scale Intermediate risk
High risk
Total
Yes
1
63
64
No
21
355
376
Total
22
418
440
Sensitivity = 63/(64) = 98 %, specificity = 21/(376) = 5.6 %, VTE venous thromboembolism
were on steroids. Previous studies have also shown that the risk of VTE is increased among steroid users in the general population [11, 12]. Experimental studies have shown that glucocorticoid use elevates the level of clotting factors and fibrinogen which subsequently increases the risk of VTE [13, 14]. Steroids are commonly prescribed to brain tumor patients with tumor progression and/or tumor edema. Therefore, steroid use likely reflects a higher tumor burden which may also increase the risk for VTE. Furthermore, the biologic causes of VTE in brain tumor patients are potentially different than other solid tumors. In contrast to other solid tumors, two recent studies have demonstrated that the concentration of circulating microparticles and expression of tissue factor on glioma cells are unrelated with VTE [15, 16]. Previous studies have shown that the long term postoperative risk to develop VTE among high grade glioma patients is 7–28 % over a 1 year period (2–4). Our results are in concordance with these prior studies. Additionally, our study reveals that patients not only develop VTEs in the initial post op period but are also at high risk to develop VTE during adjuvant chemotherapy. Moreover, we found that the majority of VTEs developed after the start of adjuvant TMZ. The fact that the risk of VTE in the post-surgical period is similar to that of the adjuvant therapy period raises the question as to whether the heightened risk for VTE in GBM patients is not related to any particular time point in the disease but instead fundamental to the disease process itself. Administration of chemotherapy has previously been associated with increased risk of VTE in cancer patients [17]. Chemotherapy can cause activation of the coagulation system through a variety of mechanisms [18–23]. Anti-angiogenic agents (such as bevacizumab) may also contribute to thrombosis through endothelial cell activation [24]. In our study, 80 % of patients who developed a VTE were on chemotherapy at the time of VTE diagnosis. This finding is similar to several other studies investigating VTE among cancer patients (9). Additionally, 47 % of the VTEs that developed after the start of adjuvant treatment were noted to occur within the first 6 months of adjuvant chemotherapy. Remarkably, there was no correlation between bevacizumab use and VTE in our study. We suspect that this
may be due to steroid requirements being reduced while on bevacizumab. A predictive model of VTE using biomarkers was recently proposed by Thaler et al. for newly diagnosed high grade glioma [25]. However, this model does not take into account any known clinical risk factors of VTE specific for this patient population. The PRODIGE trial tested the efficacy of long term anticoagulation for prevention of VTE among high grade glioma patients [26]. Unfortunately, the study had slow patient accrual. At the time of the study placebo supply expiration, the steering committee decided that continued efforts to accrue study subjects would not lead to the completion of the trial with adequate statistical power and therefore decided to close the study prematurely. In the PRODIGE study, there was a trend to reduced VTE among patients who received anticoagulation but it was not statistically significant. It remains a possibility that if only a select cohort of patients at high risk for VTE had been treated with prophylactic anticoagulation then the results could have been significant. Consequently, we feel that identifying patients at high risk to develop VTE is crucial for future trials of VTE preventive measures in GBM patients. Additionally, treatment with anticoagulation is not without risks. Anti-coagulation increases a patients risk for bleeding and an intracranial hemorrhage can potentially be life threatening. However, post-operative low molecular weight heparin is routinely used and has been found to be safe in patients with brain tumors. Low molecular weight heparin has also been shown to reduce the risk of DVT compared to compression stockings after surgical procedures [27–29]. In our study, complications of VTE treatment were observed in 16 % (n = 10) of patients. Complications included intracranial hemorrhage (4.7 %), other bleeding (4.7 %) and thrombocytopenia (6.3 %). In the PRODIGE trial, 5 % of the patients in the treatment group developed intracranial hemorrhage [26]. Developing a VTE predictive model for GBM patients may also help prevent putting patients at a lower risk for VTE on unnecessary anticoagulation. The main limitation of the study is its retrospective nature. A prospective study would likely have been more accurate in evaluating the rate of VTE in GBM patients. Additionally, since MD Anderson Cancer Center is a tertiary medical center, a significant number of the patients live far from the hospital requiring them to be followed by a local oncologist as well. VTE is usually an emergency and therefore a portion of the cases of VTE were treated locally. We assume that our patients reported all the events of VTE, but there may be cases that were not reported. Furthermore, only symptomatic VTEs were included in this study potentially causing the true number of VTEs in GBM patients to be underestimated.
123
94
J Neurooncol (2015) 124:87–94
In summary, VTE is very common in patients with GBM. However, there is no data currently supporting the use of prophylactic anticoagulation in all patients. Unfortunately, the Khorona predictive scale used for VTE risk analysis in other cancers was not validated in our study of GBM patients and an alternative scale specific for this patient population has not been developed. Creation of an accurate predictive model would allow for a focused randomized clinical trial of VTE preventive measures in high risk GBM patients and help delineate the risks to benefits of using anticoagulation for VTE prevention in this select group of high risk patients. Acknowledgments Terri S. Armstrong has received research support from Merck and Genentech. Mark R. Gilbert has received honoraria from and is on the advisory board of Merck, Genentech, and Abbott. Conflict of interest Shlomit Yust-Katz, Jacob Mandel, Jimin Wu, Courtney Webre, and Ying Yuan have no disclosures. Funding
None
References 1. Levitan N, Dowlati A, Remick SC et al (1999) Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy: risk analysis using medicare claims data. Medicine (Baltimore) 78:285–291 2. Anderson F, Huang W, Sullivan C, et al (2001) The continuing risk of venous thromboembolism following operation for glioma: findings from the Glioma Outcomes Project. Thromb Hemost, 86(Suppl):OC902 3. Brandes AA, Scelzi E, Salmistraro G et al (1997) Incidence of risk of thromboembolism during treatment high-grade gliomas: a prospective study. Eur J Cancer 33:1592–1596 4. Marras LC, Geerts WH, Perry JR (2000) The risk of venous thromboembolism is increased throughout the course of malignant glioma: an evidence based review. Cancer 89:640–646 5. Perry JR (2012) Thromboembolic disease in patients with highgrade glioma. Neuro Oncol 14(Suppl 4):iv73–iv80 6. Chaichana KL, Pendleton C, Jackson C, Quinones-Hinojosa A (2013) Deep venous thrombosis and pulmonary embolisms in adult patients undergoing craniotomy for brain tumors. Neurol Res 35(2):206–211 7. Khorana AA (2012) Risk assessment for cancer-associated thrombosis: what is the best approach? Thromb Res 129(Suppl 1):S10–S15. doi:10.1016/S0049-3848(12)70009-9 8. Khorana AA, Connolly GC (2009) Assessing risk of venous thromboembolism in the patient with cancer. J Clin Oncol 27(29):4839–4847 9. Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW (2008) Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood 111(10):4902–4907. doi:10.1182/blood-2007-10-116327 Epub 2008 Jan 23 10. Kaplan EL, Meier P (1958) Nonparametric estimation from incomplete observations. J Am Stat Assoc 53(282):457–481 11. Huerta C, Johansson S, Wallander MA, Garcıa Rodrıguez LA (2007) Risk factors and short-term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Arch Intern Med 167(9):935–943
123
12. Johannesdottir SA, Horva´th-Puho´ E, Dekkers OM et al (2013) Use of glucocorticoids and risk of venous thromboembolisma nationwide population-based case-control study. JAMA Intern Med 173(9):743–752 13. Squizzato A, Gerdes VE, Ageno W, Bu¨ller HR (2007) The coagulation system in endocrine disorders: a narrative review. Intern Emerg Med 2(2):76–83 14. van Zaane B, Nur E, Squizzato A et al (2010) Systematic review on the effect of glucocorticoid use on procoagulant, anti-coagulant and fibrinolytic factors. J Thromb Haemost 8(11):2483–2493 15. Thaler J, Ay C, Kaider A, et al. (2014) Biomarkers predictive of venous thromboembolism in patients with newly diagnosed highgrade gliomas. Neuro Oncol, nou106 (PubMed PMID: 24987133) 16. Thaler J, Preusser M, Ay C, Kaider A et al (2013) Intratumoral tissue factor expression and risk of venous thromboembolism in brain tumor patients. Thromb Res 131(2):162–165 17. Jenkins EO, Schiff D, Mackman N, Key NS (2010) Venous thromboembolism in malignant gliomas. J Thromb Hemost 8:221–227 18. Weitz IC, Israel VK, Waisman JR, Presant CA, Rochanda L, Liebman HA (2002) Chemotherapy-induced activation of hemostasis: effect of a low molecular weight heparin (dalteparin sodium) on plasma markers of hemostatic activation. Thromb Haemost 88:213–220 19. Paredes N, Xu L, Berry LR, Chan AK (2003) The effects of chemotherapeutic agents on the regulation of thrombin on cell surfaces. Br J Haematol 120:315–324 20. Walsh J, Wheeler HR, Geczy CL (1992) Modulation of tissue factor on human monocytes by cisplatin and adriamycin. Br J Haematol 81:480–488 21. Rogers JS II, Murgo AJ, Fontana JA, Raich PC (1988) Chemotherapy for breast cancer decreases plasma protein C and protein S. J Clin Oncol 6:276–281 22. Woodley-Cook J, Shin LY, Swystun L, Caruso S, Beaudin S, Liaw PC (2006) Effects of the chemotherapeutic agent doxorubicin on the protein C anticoagulant pathway. Mol Cancer Ther 5:3303–3311 23. Cwikiel M, Eskilsson J, Albertsson M, Stavenow L (1996) The influence of 5-fluorouracil and methotrexate on vascular endothelium. An experimental study using endothelial cells in the culture. Ann Oncol 7:731–737 24. Kuenen BC, Levi M, Meijers JC et al (2002) Analysis of coagulation cascade and endothelial cell activation during inhibition of vascular endothelial growth factor/vascular endothelial growth factor receptor pathway in cancer patients. Arterioscler Thromb Vasc Biol 22:1500–1505 25. Thaler J, Ay C, Kaider A, Reitter EM, Haselbo¨ck J, Mannhalter C, Zielinski C, Marosi C, Pabinger I (2014) Biomarkers predictive of venous thromboembolism in patients with newly diagnosed high-grade gliomas. Neuro Oncol, nou106 26. Perry JR, Julian JA, Laperriere NJ et al (2010) PRODIGE: a randomized placebo-controlled trial of dalteparin low-molecularweight heparin thromboprophylaxis in patients with newly diagnosed malignant glioma. J Thromb Haemost 8(9):1959–1965 27. Robins HI, O’Neill A, Gilbert M et al (2008) Effect of dalteparin and radiation on survival and thromboembolic events in glioblastoma multiforme: a phase II ECOG trial. Cancer Chemother Pharmacol 62:227–233 28. Perry SL, Bohlin C, Reardon DA et al (2009) Tinzaparin prophylaxis against venous thromboembolic complications in brain tumor patients. J Neurooncol 95:129–134 29. Salmaggi A, Simonetti G, Trevisan E, Beecher D, Carapella CM, DiMeco F, Conti L, Pace A, Filippini G (2013) Perioperative thromboprophylaxis in patients with craniotomy for brain tumours: a systematic review. J Neurooncol 113(2):293–303
CLINICAL STUDY
Venous thromboembolism (VTE) and glioblastoma Shlomit Yust-Katz1,2 • Jacob J. Mandel3 • Jimin Wu5 • Ying Yuan5 • Courtney Webre4 • Tushar A. Pawar6 • Harshad S. Lhadha6 • Mark R. Gilbert6 Terri S. Armstrong6,7
•
Received: 29 October 2014 / Accepted: 6 May 2015 / Published online: 19 May 2015 Ó Springer Science+Business Media New York 2015
Abstract The risk of venous thromboembolism (VTE) is high for patients with brain tumors (11–20 %). Glioblastoma (GBM) patients, in particular, have the highest risk of VTE (24–30 %). The Khorana scale is the most commonly used clinical scale to evaluate the risk of VTE in cancer patients but its efficacy in patients with GBM remains unclear. The aim of this study is to estimate the frequency of VTE in GBM patients and identify potential risk factors for the development of VTE during adjuvant chemotherapy. Furthermore, we intend to
Shlomit Yust-Katz and Jacob J. Mandel author have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s11060-015-1805-2) contains supplementary material, which is available to authorized users. & Shlomit Yust-Katz [email protected] 1
Neuro-Oncology Unit, Davidoff Cancer Center, Rabin Medical Center, 39 Jabotinski St., 49100 Petah Tikva, Israel
2
Sackler Faculty of Medicine, Tel Aviv University, Ramat Aviv, Tel Aviv, Israel
3
Department of Neurology, Baylor College of Medicine, 6550 Fannin Suite 1801, Houston, TX 77030, USA
4
College of Medicine, Texas A&M Health Science Center, 8447 Highway 47, Bryan, TX 778071, USA
5
Department of Biostatistics College of Medicine, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 1411, Houston, TX 77030, USA
6
Department of Neuro-Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 431, Houston, TX 77030, USA
7
Department of Family Health, The University of Texas Health Science Center-Houston SON, 6901 Bertner Ave, Houston, TX 77030, USA
examine whether the Khorana scale accurately predicts the risk of VTE in GBM patients. We retrospectively reviewed the medical records of GBM patients treated at MD Anderson during the years 2005–2011. The study cohort included 440 patients of which 64 (14.5 %) developed VTE after the start of adjuvant treatment. The median time to develop VTE was 6.5 months from the start of adjuvant treatment. On multivariate analysis male sex, BMI C 35, KPS B 80, history of VTE and steroid therapy were significantly associated with the development of VTE. The Khorana scale was found to be an invalid VTE predictive model in GBM patients due to poor specificity. Of the 64 patients who developed a VTE, 36 were treated with anticoagulation, 2 with an IVC filter, and 21 with both. Complications (intracranial hemorrhage, bleeding in other organs and thrombocytopenia) secondary to anticoagulation were reported in 16 % (n = 10). VTE is common in patients with GBM. Our results did not validate the Khorana scale in GBM patients. Additional studies identifying which GBM patients are at highest risk for VTE are needed to enable further evaluation of VTE preventive measures in this selected group. Keywords Deep vein thrombosis Pulmonary emboli Brain tumors
Introduction The risk of venous thromboembolism (VTE) is increased in patients with solid tumor malignancies [1]. Different types of solid cancers have shown varying risks for the development of VTE. Brain tumors, including glioblastoma (GBM), are one of the highest risk cancers for the development of VTE. A VTE prevalence of approximately 24–30 % (over the full trajectory of the illness) has been reported in high
123
88
J Neurooncol (2015) 124:87–94
grade glioma patients [2–4]. Several risk factors have previously been associated with the development of VTE in GBM patients including functional status, paresis, residual disease, obesity, chemotherapy and vascular endothelial growth factor targeted treatment [5]. It is also well established that the risk for VTE is high in the post surgical period and can extend up to 30 days after the surgical procedure [6]. However, it is unclear if an increased risk of VTE remains after the post-surgical period. Additionally, it is unknown if the risks associated with developing a VTE after surgery are similar to the risks of developing a VTE during adjuvant treatment with chemotherapy. Determining which patients are at a higher risk of VTE is essential as VTE occurrence can cause significant morbidity or mortality and delay vital chemotherapy. Furthermore, distinguishing patients at high risk for VTE could help improve the design of future clinical studies examining the efficacy of VTE prophylactic measures (including anticoagulation) and potentially prevent the hazard of anticoagulation in patients who do not require it [7, 8]. The most commonly used clinical scale to evaluate the risk of VTE development among cancer patients is the Khorana Chemotherapy Associated VTE Predictive Scale. The Khorana scale was developed for patients with a wide spectrum of primary cancers and includes the patient’s type of cancer, platelet count, hemoglobin level, use of erythropoietin stimulating agents, leukocyte count and body mass index(BMI) [9] (Table 1). This study estimates the frequency and examines the potential risk factors of patients with GBM who developed VTE after the start of adjuvant chemotherapy and were followed and/or treated at the MD Anderson Cancer Center. Additionally, we tested whether the Khorana scale accurately predicts the risk of VTE in GBM patients.
Materials and methods Patients After securing institutional review board approval (protocol PA12-0894), we conducted a retrospective chart review Table 1 The modified Khorana scale
of patients with primary GBM treated at MD Anderson from 2005 to 2011 and identified patients who developed a VTE during the course of their disease. All patients had a pathologically-confirmed diagnosis of primary GBM using WHO criteria. Diagnosis of VTE was determined by either ultrasound examination of the extremities, radiographic findings on CT (computed tomography) angiography of the chest or MR (magnetic resonance) venous angiography of the brain. All VTE cases identified were symptomatic (leg swelling, hand swelling, headache, dyspnea, or tachycardia). For our risk factor analysis of developing a VTE after the start of adjuvant chemotherapy, we excluded patients if they did not receive treatment with the established standard of care following surgery of concurrent chemoradiation with daily temozolomide for 6 weeks, developed a VTE before the start of adjuvant treatment or were not followed for at least 6 months after the start of adjuvant treatment. Data for the risk factor analysis (demographic and clinical characteristics) was collected on all eligible patients from the visit prior to starting the first cycle of adjuvant chemotherapy. Data regarding treatment for GBM was collected from follow-up clinic visits. Survival data was also collected from the patient’s clinical charts (date of death is documented in our institutional records). Furthermore, patients who developed a VTE after starting adjuvant treatment also had additional data collected at the time of VTE diagnosis. The additional data at time of VTE included date of VTE, type of VTE (deep vein thrombosis, pulmonary emboli, or sinus venous thrombosis), location of VTE, performance status, steroid treatment, tumor status (stable, progressing), weight, chemotherapy at time of VTE, treatment for VTE and complications of treatment. Statistical analysis Data was first summarized using standard descriptive statistics for demographic and clinical variables. Survival and time from GBM diagnosis to VTE diagnosis were evaluated using the Kaplan–Meier method and the comparison between or among patients’ characteristics was assessed using log-rank test [10]. Both univariate and multivariate Cox regression models were applied to assess
Patient characteristics
Risk score
Site of cancer Very high risk (brain, stomach, pancreas) High risk (lung, lymphoma, gynecologic, bladder, testicular)
1
Pre-chemotherapy platelet count C350,000/mm3
1
Hemoglobin level \10 g/dl or use of red cell growth factors
1
Pre-chemotherapy leukocyte count [11,000/mm3
1
2
Body mass index C35 kg/m
High-risk score C 3; Intermediate-risk score = 1–2; Low-risk score = 0
123
2
1
J Neurooncol (2015) 124:87–94
the effect of covariates of interest on the risk to develop VTE. All statistical analyses were carried out in SAS 9.3 (SAS Institute Inc., Cary, NC, USA). The Khorana Chemotherapy Associated VTE Predictive Scale predicts the risk of developing VTE in cancer patients [7–9]. It is assessed according to several risk factors (see Table 1). We applied the modified Khorana scale that includes high, intermediate, and low risk groups to the collected data in order to classify patients into those groups and then calculated its sensitivity and specificity. Khorana et al. reported that the Khorana scale has a sensitivity of 40.0 % and a specificity of 88 % based on a validation cohort of 1365 patients [9]. If the sensitivity and specificity of the Khorana scale in our collected data is comparable to these reported values then we will regard the Khorana scale applicable to the GBM patient population.
Results A total of 543 patients with primary GBM treated at MD Anderson during the years 2005–2011 were identified. Of these patients, 117 (22 %) were diagnosed with VTE during the course of their illness. 64 (55 %) patients developed VTE after starting adjuvant chemotherapy and 53 (45 %) developed VTE before starting adjuvant treatment [20 (17 %) during the post operative period and 33 (28 %) during radiation or during the time between radiation and the start of adjuvant treatment]. The target population for the risk factor analysis of our study was patients who developed VTE after starting adjuvant chemotherapy. Therefore, in our analysis, we excluded patients that developed VTE before starting adjuvant treatment (53 patients), as well as patients who did not receive the current standard of care with concurrent radiation and temozolomide (TMZ) prior to starting adjuvant chemotherapy treatment or were not followed for at least 6 months after starting adjuvant treatment (50 patients). Thus, the study sample included 440 patients and 64 (14.5 %) of them developed a VTE. The clinical characteristics of our study population are summarized in Table 2 and in the supplementary data (Table 1s). Patient’s characteristics at time of VTE diagnosis Sixty four patients developed VTE after initiation of adjuvant chemotherapy. Their characteristics are described in in the supplementary data (Tables 2s and 3s). 15 patients developed pulmonary emboli (PE), 27 had deep vein thrombosis (DVT), 20 developed both a DVT and PE, and sinus vein thrombosis was diagnosed in 2 patients. Of the 47 patients who developed a DVT, two (4 %) developed a DVT in the upper extremities. Both of these
89
patients had central lines and the DVTs occurred on the same side as the central line. Median age at time of VTE was 56 (range 19–80). The median time to develop VTE was 6.5 months (Range: 0.03–120.74). Thirty (46.9 %) patients developed VTE within 6 months after starting adjuvant chemotherapy. 60 % (37) of patients were ambulatory without an assistive device at time of VTE diagnosis and 30 % (19) of the patients had evidence of hemiparesis on exam. The median karnofsky performance score (KPS) at time of VTE diagnosis was 80 (20–90). The majority of patients with VTE [72.6 % (45)] were on steroids at time of VTE diagnosis. Eighty one percent (52) of patients were being treated with chemotherapy at the time of VTE diagnosis. 48 % (25) were on a TMZ containing regimen, 23 % (12) on a bevacizumab containing regimen, 4 % (2) on both TMZ and bevacizumab, and 25 % (13) on various other chemotherapies. There was evidence of clinical or radiographic progression in 48 % (31) of patients at time of VTE diagnosis. Three patients (4.7 %) were classified later as having had pseudoprogression. Of the 64 patients who developed a VTE, 36 were treated with anticoagulation alone (8 patients with Coumadin and the rest with low molecular weight heparin), 2 had an inferior vena-cava (IVC) filter alone, and 21 patients received both an IVC filter and anticoagulation. Treatment was unknown for 3 patients and another 2 patients were not treated as they were referred to hospice. 40 % of patients with a VTE were admitted for treatment. Most of the patients (93 %) receiving anticoagulation were treated for greater than 6 months. The decision to insert an IVC filter at time of VTE diagnosis was determined by the treating physicians preference based upon history of a hemorrhagic event prior to VTE diagnosis, development of a complication while on anti-coagulation, or desire to delay anticoagulation due to a recent or upcoming surgical resection. Complications secondary to anticoagulation were reported in 16 % (n = 10) of patients. These complications included brain hemorrhage in 3 patients (subdural and intracerebral), bleeding to other organs in 3 patients (rectal sheath, rectal and bleeding to Port- A-Cath), and thrombocytopenia in 4 patients (one who was diagnosed with heparin induced thrombocytopenia). Four of the patients that developed complications secondary to treatment (two with intracranial bleeding, one with a rectal sheath hematoma and one with heparin induced thrombocytopenia) were sent to hospice after admission for the complication. Only 2.5 % (n = 11) of patients had a history of VTE before diagnosis. A very small minority of the patients had a history of stroke (1.1 % n = 5), myocardial infarct (2 %
123
90
J Neurooncol (2015) 124:87–94
Table 2 Summary of demographic and clinical characteristic of patients diagnosed with venous thromboembolism at time of venous thromboembolism diagnosis -categorical covariate Covariate
Levels
Patient diagnosed with VTE Number of events Location of VTE
Pulmonary embolus (PE) Number of brain surgeries before VTE diagnosis
Number of hospitalizations in the last 6 months
Steroid therapy
Tumor status at the time of thrombosis diagnosis
KPS
Paresis
Ambulatory
BMI
Chemotherapy at time of VTE
123
N (%) 64 (100 %)
1
63 (98 %)
2
1 (1.6 %)
One leg
38 (59 %)
Two legs
7 (11 %)
Hand
2 (3.2 %)
PE with no proven evidence of DVT Sinus vein thrombosis
15 (23.8 %) 2 (3.2 %)
No
29 (45.3 %)
Yes
35 (54.7 %)
1
37 (57.8 %)
2
20 (31.3 %)
3 and more
7 (11 %)
Unknown
2
0
16 (25.8 %)
1
34 (54.8 %)
2 or more
12 (19 %)
Unknown
3
No
16 (25.8 %)
Yes
45 (73.6 %)
Stable
26 (40.6 %)
Radiographic progression only Radiographic progression and clinical progression
11 (17.2 %) 23 (35.9 %)
Clinical progression
4 (6.3 %)
Unknown
9
B80
42 (80 %)
90
13 (20 %)
Unknown
2
None
33 (53.2 %)
Upper limb paresis or plegia
3 (4.8 %)
Lower Limb Paresis
2 (3.2 %)
Hemiparesis
19 (30.6 %)
Paraplegia
1 (1.6 %)
Hemiplegia
3 (4.8 %)
Quadriplegia
1 (1.6 %)
Unknown
3
No
13 (21.3 %)
Ambulatory without aid Ambulatory with walker
37 (60.7 %) 11 (18 %)
Unknown
7
\35
45 (78.9 %)
C35
12 (21.1 %)
Unknown
2
None
10 (16.1 %)
Bev alone
1(1.6 %)
Bev plus other chemotherapy
11 (17.7 %)
Tmz alone
10 (16.1 %)
Tmz plus other chemotherapy
11 (17.7 %)
J Neurooncol (2015) 124:87–94
91
Table 2 continued Covariate
Central venous catheter
Treatment for VTE
Duration of treatment
Complications of treatment
Admitted for VTE
Levels
N (%)
Tmz plus thalidomide regimen
4 (6.5 %)
Tmz plus bev
2 (3.2 %)
Other
13 (21 %)
Unknown
2
No
57 (91.9 %)
Yes
5 (8.1 %)
Unknown
3
Anticoagulation only
36 (59 %)
IVC filter only
2 (3 %)
Anticoagulation and IVC filter
21 (34 %)
Hospice and no treatment Unknown
2 (3 %) 21
6 Months or more
40 (93 %)
1 Month or less
3 (7 %)
No complications
54 (84.3 %)
Thrombocytopenia
4 (6.3 %)
Brain hemorrhage
3 (4.7 %)
Bleeding other
3(4.7 %)
Unknown
22
No
25 (59.5 %)
Yes
17 (40.5 %)
VTE venous thromboembolism, PE pulmonary emboli, BMI body mass index, KPS karnofsky performance status
n = 9) or atrial fibrillation (1.4 %, n = 6). History of smoking was positive in 8.9 % (n = 39) of the patients. 90 % (n = 393) of patients were able to ambulate without assistance at the time of starting adjuvant treatment. Only 34 % (145) of patients were on steroids at the time of starting adjuvant treatment. A complete description of the concomitant medication that patients were on at the time of starting adjuvant treatment is listed in the supplementary data (Table 1s). A variety of dosing regimens were used for adjuvant TMZ. Most patients (60 %, n = 243) were treated with the Stupp regimen (5/28). At the start of adjuvant chemotherapy, a high percentage of patients (53.2 %, n = 218) were treated with other chemotherapeutic agents in addition to adjuvant TMZ usually as part of clinical trials. 67 % (n = 295) of patients were treated with bevacizumab during the course of their illness. A small percentage (6.8 %, n = 30) of patients were not treated with adjuvant TMZ due to either tumor progression after chemoradiation or low blood counts. Some of these patients were treated with other adjuvant treatments after chemoradiation. Several patients (3.9 %, n = 17) were being treated with anticoagulation at the time of GBM diagnosis for a variety of reasons including history of VTE and atrial fibrillation. An additional 25 patients (5.7 %) were started on 1 mg of coumadin prophylactically at the time of
adjuvant therapy as part of a clinical trial containing thalidomide. Risk factors for the development of VTE We next assessed which factors were associated with risk to develop VTE by univariate analysis. The significant categorical covariates were sex (p = 0.0102), history of venous thrombosis (p = 0.0056), KPS at start of adjuvant TMZ (p = 0.0316), paresis (p = 0.0004), ambulatory status (p = 0.0138), and BMI (p = 0.0047). Concomitant medications that were associated with risk of VTE included anticoagulation (p = 0.0105), anti-hypertensives (p = 0.0264), and steroid therapy (p \ 0.0001). Anti- gastroesophageal reflux drugs (p = 0.0106) correlated with the risk of VTE but were also found to be associated with corticosteroid use. The significant continuous covariates were age at diagnosis (p = 0.0448), KPS at start of adjuvant TMZ (p = 0.0028) (KPS was assessed as continuous and then categorized), pre-adjuvant TMZ white blood cell count (WBC) (p = 0.0010), pre- adjuvant TMZ absolute neutrophil count (ANC) (p = 0.0001), and steroid dose (p = 0.0014). Factors that were significant (p value \ 0.05) in the univariate cox regression analysis were then taken into a
123
92
J Neurooncol (2015) 124:87–94
multivariate cox regression model. On multivariate analysis, sex (p \ 0.0122), BMI (p \ 0.0145), KPS at start of adjuvant TMZ (p \ 0.0125), history of VTE (p \ 0.0031) and steroid therapy (p \ 0.0115) remained significant (Table 3). Khorana scale We then evaluated whether the Khorana scale is predictive for thrombosis in the GBM population. Using the Khorana scale, all GBM patients were assigned to the very high risk category (risk score = 2) in terms of the cancer site and therefore required only 1 additional risk factor to reach the total risk score C3 to be considered as high-risk overall. As a consequence, most of the patients (n = 418) were categorized as high risk when the scale was applied (Table 4). 63/64 patients that developed VTE were in the high risk group for a sensitivity of 98 %. However, only 21/376 patients that did not develop VTE were in the intermediate risk group for a specificity of 5.6 %.
Discussion This study confirms the high prevalence of VTE in GBM patients with 22 % of GBM patients diagnosed with a VTE. In addition, we found that VTEs commonly occur during adjuvant chemotherapy (64, 55 %) with similar incidence to the postoperative period (53, 45 %). Our study confirms that obesity, history of VTE, elevated WBCs and low ANCs correlate with a higher risk of VTE in GBM
patients. Additionally, several factors that are more specific for GBM patients including functional status and steroid treatment were also found to correlate with VTE risk. The Khorana scale was found to be lacking adequate specificity to predict the risk of VTE in this patient population and therefore is not an adequate tool to predict the risk of VTE in GBM patients. The Khorana scale may not be valid in GBM patients for several reasons. Perhaps, due to the high risk associated with the GBM diagnosis alone, the importance of other specific risk factors are not accurately differentiated or weighed. Another possibility is that the factors included in the Khorana scale may not be relevant in the GBM population. Distinct from other cancers, brain tumor patients often have a poor performance status and increased steroid requirements, both of which have been associated with increased VTE risk [5]. Notably, the dataset from which the Khorana scale was developed included only a few brain tumor patients (n = 4) and patients with poor performance status were underrepresented [9]. This study is one of the first to evaluate the Khorana scale utility in this high risk patient population. A variety of clinical risk factors including age, obesity, medical comorbidity, prior episodes of thrombosis, and hematological parameters, have been associated with development of VTE among cancer patients [8]. Our study confirmed that obesity, history of VTE, elevated WBCs, and low ANCs correlate with a higher risk of VTE in GBM patients. Additionally, several factors that are more specific for GBM patients, including functional status and steroid treatment were also found to correlate with VTE risk. Notably, 72.6 % of GBM patients who developed VTE
Table 3 Multivariate Cox regression model Parameter
Level
Hazard ratio
95 % Confidence limits
p value
Sex
Male versus female
2.33
1.20
4.50
0.0122
History of VTE
Yes versus No
4.50
1.66
12.19
0.0031
KPS at start of adjuvant TMZ
50–80 versus 90–100
2.22
1.19
4.15
0.0125
Paresis BMI
Others versus none C35 versus \35
1.18 2.65
0.54 1.21
2.58 5.80
0.6841 0.0145
Ambulatory status
Others versus none
0.73
0.09
6.02
0.7705
Blood pressure medication
Yes versus No
1.06
0.56
2.03
0.8564
GERD medication
Yes versus No
1.08
0.59
1.98
0.7927
Steroid therapy
Yes versus No
2.20
1.19
4.05
0.0115
Age at diagnosis
1.01
0.98
1.03
0.6845
WBC at start of adjuvant TMZ
0.91
0.67
1.23
0.5385
ANC at start of adjuvant TMZ
1.24
0.90
1.71
0.1942
VTE venous thromboembolism, KPS karnofsky performance status, BMI body mass index, GERD gastroesophageal reflux disease, WBC white blood cell count, ANC absolute neutrophil count
123
J Neurooncol (2015) 124:87–94
93
Table 4 Applying the Khorana scale for the study population VTE
Khorana scale Intermediate risk
High risk
Total
Yes
1
63
64
No
21
355
376
Total
22
418
440
Sensitivity = 63/(64) = 98 %, specificity = 21/(376) = 5.6 %, VTE venous thromboembolism
were on steroids. Previous studies have also shown that the risk of VTE is increased among steroid users in the general population [11, 12]. Experimental studies have shown that glucocorticoid use elevates the level of clotting factors and fibrinogen which subsequently increases the risk of VTE [13, 14]. Steroids are commonly prescribed to brain tumor patients with tumor progression and/or tumor edema. Therefore, steroid use likely reflects a higher tumor burden which may also increase the risk for VTE. Furthermore, the biologic causes of VTE in brain tumor patients are potentially different than other solid tumors. In contrast to other solid tumors, two recent studies have demonstrated that the concentration of circulating microparticles and expression of tissue factor on glioma cells are unrelated with VTE [15, 16]. Previous studies have shown that the long term postoperative risk to develop VTE among high grade glioma patients is 7–28 % over a 1 year period (2–4). Our results are in concordance with these prior studies. Additionally, our study reveals that patients not only develop VTEs in the initial post op period but are also at high risk to develop VTE during adjuvant chemotherapy. Moreover, we found that the majority of VTEs developed after the start of adjuvant TMZ. The fact that the risk of VTE in the post-surgical period is similar to that of the adjuvant therapy period raises the question as to whether the heightened risk for VTE in GBM patients is not related to any particular time point in the disease but instead fundamental to the disease process itself. Administration of chemotherapy has previously been associated with increased risk of VTE in cancer patients [17]. Chemotherapy can cause activation of the coagulation system through a variety of mechanisms [18–23]. Anti-angiogenic agents (such as bevacizumab) may also contribute to thrombosis through endothelial cell activation [24]. In our study, 80 % of patients who developed a VTE were on chemotherapy at the time of VTE diagnosis. This finding is similar to several other studies investigating VTE among cancer patients (9). Additionally, 47 % of the VTEs that developed after the start of adjuvant treatment were noted to occur within the first 6 months of adjuvant chemotherapy. Remarkably, there was no correlation between bevacizumab use and VTE in our study. We suspect that this
may be due to steroid requirements being reduced while on bevacizumab. A predictive model of VTE using biomarkers was recently proposed by Thaler et al. for newly diagnosed high grade glioma [25]. However, this model does not take into account any known clinical risk factors of VTE specific for this patient population. The PRODIGE trial tested the efficacy of long term anticoagulation for prevention of VTE among high grade glioma patients [26]. Unfortunately, the study had slow patient accrual. At the time of the study placebo supply expiration, the steering committee decided that continued efforts to accrue study subjects would not lead to the completion of the trial with adequate statistical power and therefore decided to close the study prematurely. In the PRODIGE study, there was a trend to reduced VTE among patients who received anticoagulation but it was not statistically significant. It remains a possibility that if only a select cohort of patients at high risk for VTE had been treated with prophylactic anticoagulation then the results could have been significant. Consequently, we feel that identifying patients at high risk to develop VTE is crucial for future trials of VTE preventive measures in GBM patients. Additionally, treatment with anticoagulation is not without risks. Anti-coagulation increases a patients risk for bleeding and an intracranial hemorrhage can potentially be life threatening. However, post-operative low molecular weight heparin is routinely used and has been found to be safe in patients with brain tumors. Low molecular weight heparin has also been shown to reduce the risk of DVT compared to compression stockings after surgical procedures [27–29]. In our study, complications of VTE treatment were observed in 16 % (n = 10) of patients. Complications included intracranial hemorrhage (4.7 %), other bleeding (4.7 %) and thrombocytopenia (6.3 %). In the PRODIGE trial, 5 % of the patients in the treatment group developed intracranial hemorrhage [26]. Developing a VTE predictive model for GBM patients may also help prevent putting patients at a lower risk for VTE on unnecessary anticoagulation. The main limitation of the study is its retrospective nature. A prospective study would likely have been more accurate in evaluating the rate of VTE in GBM patients. Additionally, since MD Anderson Cancer Center is a tertiary medical center, a significant number of the patients live far from the hospital requiring them to be followed by a local oncologist as well. VTE is usually an emergency and therefore a portion of the cases of VTE were treated locally. We assume that our patients reported all the events of VTE, but there may be cases that were not reported. Furthermore, only symptomatic VTEs were included in this study potentially causing the true number of VTEs in GBM patients to be underestimated.
123
94
J Neurooncol (2015) 124:87–94
In summary, VTE is very common in patients with GBM. However, there is no data currently supporting the use of prophylactic anticoagulation in all patients. Unfortunately, the Khorona predictive scale used for VTE risk analysis in other cancers was not validated in our study of GBM patients and an alternative scale specific for this patient population has not been developed. Creation of an accurate predictive model would allow for a focused randomized clinical trial of VTE preventive measures in high risk GBM patients and help delineate the risks to benefits of using anticoagulation for VTE prevention in this select group of high risk patients. Acknowledgments Terri S. Armstrong has received research support from Merck and Genentech. Mark R. Gilbert has received honoraria from and is on the advisory board of Merck, Genentech, and Abbott. Conflict of interest Shlomit Yust-Katz, Jacob Mandel, Jimin Wu, Courtney Webre, and Ying Yuan have no disclosures. Funding
None
References 1. Levitan N, Dowlati A, Remick SC et al (1999) Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy: risk analysis using medicare claims data. Medicine (Baltimore) 78:285–291 2. Anderson F, Huang W, Sullivan C, et al (2001) The continuing risk of venous thromboembolism following operation for glioma: findings from the Glioma Outcomes Project. Thromb Hemost, 86(Suppl):OC902 3. Brandes AA, Scelzi E, Salmistraro G et al (1997) Incidence of risk of thromboembolism during treatment high-grade gliomas: a prospective study. Eur J Cancer 33:1592–1596 4. Marras LC, Geerts WH, Perry JR (2000) The risk of venous thromboembolism is increased throughout the course of malignant glioma: an evidence based review. Cancer 89:640–646 5. Perry JR (2012) Thromboembolic disease in patients with highgrade glioma. Neuro Oncol 14(Suppl 4):iv73–iv80 6. Chaichana KL, Pendleton C, Jackson C, Quinones-Hinojosa A (2013) Deep venous thrombosis and pulmonary embolisms in adult patients undergoing craniotomy for brain tumors. Neurol Res 35(2):206–211 7. Khorana AA (2012) Risk assessment for cancer-associated thrombosis: what is the best approach? Thromb Res 129(Suppl 1):S10–S15. doi:10.1016/S0049-3848(12)70009-9 8. Khorana AA, Connolly GC (2009) Assessing risk of venous thromboembolism in the patient with cancer. J Clin Oncol 27(29):4839–4847 9. Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW (2008) Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood 111(10):4902–4907. doi:10.1182/blood-2007-10-116327 Epub 2008 Jan 23 10. Kaplan EL, Meier P (1958) Nonparametric estimation from incomplete observations. J Am Stat Assoc 53(282):457–481 11. Huerta C, Johansson S, Wallander MA, Garcıa Rodrıguez LA (2007) Risk factors and short-term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Arch Intern Med 167(9):935–943
123
12. Johannesdottir SA, Horva´th-Puho´ E, Dekkers OM et al (2013) Use of glucocorticoids and risk of venous thromboembolisma nationwide population-based case-control study. JAMA Intern Med 173(9):743–752 13. Squizzato A, Gerdes VE, Ageno W, Bu¨ller HR (2007) The coagulation system in endocrine disorders: a narrative review. Intern Emerg Med 2(2):76–83 14. van Zaane B, Nur E, Squizzato A et al (2010) Systematic review on the effect of glucocorticoid use on procoagulant, anti-coagulant and fibrinolytic factors. J Thromb Haemost 8(11):2483–2493 15. Thaler J, Ay C, Kaider A, et al. (2014) Biomarkers predictive of venous thromboembolism in patients with newly diagnosed highgrade gliomas. Neuro Oncol, nou106 (PubMed PMID: 24987133) 16. Thaler J, Preusser M, Ay C, Kaider A et al (2013) Intratumoral tissue factor expression and risk of venous thromboembolism in brain tumor patients. Thromb Res 131(2):162–165 17. Jenkins EO, Schiff D, Mackman N, Key NS (2010) Venous thromboembolism in malignant gliomas. J Thromb Hemost 8:221–227 18. Weitz IC, Israel VK, Waisman JR, Presant CA, Rochanda L, Liebman HA (2002) Chemotherapy-induced activation of hemostasis: effect of a low molecular weight heparin (dalteparin sodium) on plasma markers of hemostatic activation. Thromb Haemost 88:213–220 19. Paredes N, Xu L, Berry LR, Chan AK (2003) The effects of chemotherapeutic agents on the regulation of thrombin on cell surfaces. Br J Haematol 120:315–324 20. Walsh J, Wheeler HR, Geczy CL (1992) Modulation of tissue factor on human monocytes by cisplatin and adriamycin. Br J Haematol 81:480–488 21. Rogers JS II, Murgo AJ, Fontana JA, Raich PC (1988) Chemotherapy for breast cancer decreases plasma protein C and protein S. J Clin Oncol 6:276–281 22. Woodley-Cook J, Shin LY, Swystun L, Caruso S, Beaudin S, Liaw PC (2006) Effects of the chemotherapeutic agent doxorubicin on the protein C anticoagulant pathway. Mol Cancer Ther 5:3303–3311 23. Cwikiel M, Eskilsson J, Albertsson M, Stavenow L (1996) The influence of 5-fluorouracil and methotrexate on vascular endothelium. An experimental study using endothelial cells in the culture. Ann Oncol 7:731–737 24. Kuenen BC, Levi M, Meijers JC et al (2002) Analysis of coagulation cascade and endothelial cell activation during inhibition of vascular endothelial growth factor/vascular endothelial growth factor receptor pathway in cancer patients. Arterioscler Thromb Vasc Biol 22:1500–1505 25. Thaler J, Ay C, Kaider A, Reitter EM, Haselbo¨ck J, Mannhalter C, Zielinski C, Marosi C, Pabinger I (2014) Biomarkers predictive of venous thromboembolism in patients with newly diagnosed high-grade gliomas. Neuro Oncol, nou106 26. Perry JR, Julian JA, Laperriere NJ et al (2010) PRODIGE: a randomized placebo-controlled trial of dalteparin low-molecularweight heparin thromboprophylaxis in patients with newly diagnosed malignant glioma. J Thromb Haemost 8(9):1959–1965 27. Robins HI, O’Neill A, Gilbert M et al (2008) Effect of dalteparin and radiation on survival and thromboembolic events in glioblastoma multiforme: a phase II ECOG trial. Cancer Chemother Pharmacol 62:227–233 28. Perry SL, Bohlin C, Reardon DA et al (2009) Tinzaparin prophylaxis against venous thromboembolic complications in brain tumor patients. J Neurooncol 95:129–134 29. Salmaggi A, Simonetti G, Trevisan E, Beecher D, Carapella CM, DiMeco F, Conti L, Pace A, Filippini G (2013) Perioperative thromboprophylaxis in patients with craniotomy for brain tumours: a systematic review. J Neurooncol 113(2):293–303
