Keywords
glioblastoma - circulating tumor cells - deep vein thrombosis - myocardial infarction
Introduction
Glioblastoma (GBM) is one of the most deadly cancer entities, with a 5-year survival rate of 5%.[1] Patients with GBM are at high risk of developing venous and arterial thromboembolism (VTE, ATE). Cancer-associated thrombosis (CAT) occurs in up to 20 to 30% of GBM patients,[2]
[3]
[4]
[5] and significantly contributes to morbidity and mortality.[4] Known risk factors for venous and arterial thromboembolism (TE) in GBM patients include performance of brain surgery, the use of corticosteroids as well as the use of radiation and chemotherapy.[5]
[6] Additionally, several tumor-type specific risk factors such as high expression of procoagulant tissue factor (TF) and podoplanin on cancer cells have been identified to promote a procoagulant state in GBM.[7]
[8] Intriguingly, recent studies demonstrated that circulating tumor cells (CTCs) are present in the blood from patients with GBM.[9]
[10]
[11]
[12]
[13]
Of note, cancer cells were also present in thrombi specimen of cancer patients who had developed VTE, indicating that CTCs directly promote CAT.[14] It is therefore very tempting to speculate that CTCs with procoagulant properties derived from the primary tumor trigger coagulation activation in GBM, once they come in contact with the hemostatic system. However, while in vitro findings support that tumor cells can initiate coagulation and platelet aggregation, there is a substantial lack of clinical data.
The aim of our study was to assess whether detection of CTCs predicts an increased risk of TE in GBM.
Materials and Methods
Study Design
This retrospective post hoc study was conducted at the University Medical Center Hamburg-Eppendorf, Germany. Patients with GBM who underwent treatment in the Department of Neurological Surgery from May 2005 to August 2013 and who had been investigated for CTCs in a previous study within the ERC-2010-AdG_20100317 DISSECT project[9] were analyzed for documented arterial and venous thromboembolic events at the time point of or following diagnosis of GBM. CTCs were detected as previously described.[9] Data were retrieved from electronic patient records and included demographic and clinical patient characteristics (gender, age at the time of diagnosis of GBM, medication including antithrombotic therapy) and laboratory parameters (complete blood count, global coagulation tests) on the day of admission, and thromboembolic adverse events. Eight patients from the original cohort had to be excluded since their clinical data were not available in our current electronic chart system. Follow-up electronic records were assessed until December 2022.
VTE comprised deep vein thrombosis (DVT) and/or pulmonary embolism (PE). ATE was defined as transitory ischemic attack, ischemic stroke, ST-elevation myocardial infarction, non–ST-elevation myocardial infarction (NSTEMI), and systemic arterial embolism. TE events which are clearly linked to another thromboembolic risk factor (e.g., atrial fibrillation) were excluded.
Statistical Analysis
Statistical analyses were performed using RStudio, version 3.6.1. Patient characteristics were reported by descriptive statistics. Quantitative variables were summarized as median with interquartile range (IQR). Differences between groups were compared using the Mann–Whitney U-test. Differences between categorical data were analyzed using the chi-square test. A p-value <0.05 was considered statistically significant.
Results
We analyzed 133 patients; their median age was 63 years (IQR, 53–70 years) and 56 (42%) were women. Twenty-three patients (17%) had a recurrent GBM. CTCs were detected in 26 patients (20%). Median follow-up time was 7.7 months. Fourteen patients (11%) experienced thromboembolic complications, involving 8 (6%) VTEs and 6 (5%) ATEs ([Table 1]).
Table 1
Demographic data
|
N
|
133
|
|
Age in years, median (IQR)
|
63 (53–70)
|
|
Female sex, n (%)
|
56 (42.1)
|
|
Recurrent GBM, n (%)
|
23 (17.3)
|
|
CTCs, n (%)
|
26 (19.5)
|
|
Hemoglobin in g/L, median (IQR)
|
142 (135–151)
|
|
Leukocytes in 109/L, median (IQR)
|
10 (7.6–13.3)
|
|
Platelets in 109/L, median (IQR)
|
257 (219–295)
|
|
aPTT in sec, median (IQR)
|
27.2 (24.10–29.80)
|
|
INR, median (IQR)
|
1 (0.97–1.05)
|
|
BMI in kg/m2, median (IQR)
|
24.8 (22.9–27.7)
|
|
Comorbidities, n (%)
|
|
|
● Arterial hypertension
● Hyperlipidemia
● Diabetes mellitus
● Cardiovascular disease
● Smoking
● History of stroke or TIA
● Obesity
● History of VTE
● Peripheral artery disease
|
● 44 (33)
● 13 (9.8)
● 12 (9.0)
● 10 (7.5)
● 9 (6.8)
● 6 (4.5)
● 5 (3.8)
● 3 (2.3)
● 2 (1.5)
|
|
Thromboembolic complications, n (%)
|
14 (11)
|
|
● VTE
- DVT[a]
- PE[a]
● ATE
- NSTEMI
- Ischemic stroke
- Arterial embolism
|
● 8 (6.0)
- 5 (3.8)
- 6 (4.5)
● 6 (4.5)
- 2 (1.5)
- 3 (2.3)
- 1 (0.8)
|
Abbreviations: aPTT, activated partial thromboplastin time; ATE, arterial thromboembolism; BMI, body mass index; CTCs, circulating tumor cells; DVT, deep vein thrombosis; GBM, glioblastoma; INR, international normalized ratio; NSTEMI, non–ST-elevation myocardial infarction; PE, pulmonary embolism; TIA, transitory ischemic attack; VTE, venous thromboembolism.
Note: Data are presented as n (%) unless otherwise specified.
a Including more than one event per patient.
VTE comprised PE in six patients and DVT in five patients including three patients having both manifestations. Among ATEs, there were two cases of NSTEMI, three cases of ischemic stroke, and one case of arterial embolism in the absence of atrial fibrillation. All events occurred either within 2 months prior to diagnosis (two VTEs) or during follow-up. Our findings confirm that diagnosis of GBM is associated with a high incidence of arterial and venous thromboembolic complications. Among patients with CTCs, 4 out of 26 (15%) developed TE compared with 10 out of 107 (9%) patients without detectable CTCs (chi-square p = 0.59, [Table 2]). Of these, VTE occurred in 2 out of 26 (8%) patients with CTCs compared with 6 out of 107 (6%) patients without CTCs. There was also no difference in the incidence of ATEs, which occurred in two patients with CTCs compared with four patients without CTCs (8 vs. 4%). All four thromboembolic events in patients with CTCs occurred after surgery.
Table 2
Study population stratified by TE
|
TE positive
|
TE negative
|
p-Value
|
|
n
|
14
|
119
|
|
|
CTCs, n (%)
|
4 (28.6)
|
22 (18.5)
|
0.59
|
|
Age in years, median (IQR)
|
69 (51–73)
|
63 (53–69)
|
0.39
|
|
Female sex, n (%)
|
6 (42.9)
|
50 (42)
|
1.00
|
|
Recurrent GBM, n (%)
|
2 (14.3)
|
21 (18.1)
|
1.00
|
|
Hemoglobin in g/L, median (IQR)
|
146 (139–152)
|
142 (135–150)
|
0.277
|
|
Leukocytes in 109/L, median (IQR)
|
12.4 (9.17–15.92)
|
9.90 (7.30–13.20)
|
0.082
|
|
Platelets in 109/L, median (IQR)
|
287 (223.5–378.75)
|
253 (219–289)
|
0.119
|
|
aPTT in sec, median (IQR)
|
29.10 (25.75–30.83)
|
27.10 (23.90–29.50)
|
0.194
|
|
INR, median (IQR)
|
1.04 (0.99–1.06)
|
1.00 (0.97–1.03)
|
0.048
|
|
BMI in kg/m2, median (IQR)
|
26.75 (24.95–27.65)
|
24.55 (22.50–27.55)
|
0.161
|
Abbreviations: aPTT, activated partial thromboplastin time; BMI, body mass index; CTCs, circulating tumor cells; GBM, glioblastoma; INR, international normalized ratio; TE, thromboembolism.
Note: Data are presented as n (%) unless otherwise specified.
Next, we investigated whether other laboratory and clinical risk factors previously associated with CAT contributed to thromboembolic events in our patient cohort ([Table 2]). Age, body mass index (BMI), hemoglobin (Hb), or platelet counts did not differ between patients with documented TE and those without TE. Patients with TE had a higher international normalized ratio (INR) at baseline: 1.04 (0.99–1.06) versus 1.00 (0.97–1.03), p = 0.048. Also, there was modest evidence toward a higher leukocyte count: 12.4 × 109/L (9.17–15.92 × 109/L) versus 9.9 × 109/L (7.30–13.20 × 109/L, p = 0.08) in patients with TE compared with patients without TE which is supported by previous findings.[15]
We did not see a major difference in median follow-up times between patients with TE and those without TE (7.4 vs. 7.8 months, p = 1).
In summary, the presence of CTCs is not associated with an enhanced risk of TE in our cohort of patients with GBM. Among clinical and laboratory risk factors for CAT,[16] leukocyte counts were higher in patients who developed TE.
Discussion
In our study, we did not find an association of CTCs with thromboembolic complications in patients with GBM, a type of cancer that is characterized by a very high risk of TE.[4] GBM is a highly vascularized tumor, frequently found with intravascular microthrombosis[17] and upregulation of TF and platelet-activating podoplanin on the surface of tumor cells.[18]
[19] While GBM rarely metastasizes extracranially,[20] which might be masked by the short life expectancy following diagnosis, the TE risk is similar to that of other solid cancer patients with advanced, metastatic disease,[15]
[21] pointing toward an underlying systemic hypercoagulable state in this primarily localized disease setting.
Data on a possible involvement of CTCs in CAT are scarce and do not exist for GBM. Importantly, CTCs indicate a (subsequent) metastatic disease state in various solid cancers and correlate with the number of metastases.[22] Accordingly, high levels of procoagulant CTCs may contribute to the greater VTE risk observed in patients with advanced, disseminated cancer stage.[23]
A study by Gi et al[14] sheds some light on a possible involvement of CTCs in CAT: cancer cells were found in 27% of thrombi specimen of autopsied cancer patients (involving lung, pancreas, and gastrointestinal cancers) with VTE. Here, cancer cells were either in small cell clusters, fully surrounded by thrombus material, or were directly invading endothelial cells within the thrombus. Importantly, up to 88% of all thrombi specimens contained cancer cells that stained positive for TF or podoplanin. These findings indicate that CTCs—either directly via surface-expressed TF or podoplanin, and/or indirectly via endothelial cell activation—may induce coagulation activation, leading to VTE in cancer patients.
So far, an association of TE and CTCs has been found only in disseminated, advanced disease stage: metastatic breast cancer patients tested positive for CTCs had significantly higher TE incidence rates compared with CTC-negative patients as shown in two different studies.[24]
[25] It has to be noted, however, that an increased VTE risk was also linked to a high number of metastases, with a hazard ratio of 8.08 (p = 0.001) compared to a hazard ratio of 5.29 for CTCs (p = 0.007).[24] In another post hoc analysis of metastatic breast cancer patients, the TE incidence rate was 13 per 100 patient-years in patients with ≥ 1 CTCs compared to a TE incidence rate of 4 per 100 patient-years in patients without CTCs. Here, detection of CTCs was confirmed as an independent risk factor by multivariate analysis.[25] Also, in metastatic breast cancer, the number of CTCs correlated with plasma levels of D-dimer, fibrinogen, and thrombin–antithrombin complexes, indicating global coagulation activation.[26]
[27] Functional analyses are warranted to delineate whether CTCs not only constitute a surrogate biomarker for advanced, disseminated disease state associated with an elevated TE risk but are also the prothrombotic driver themselves.
In line with our findings, CTCs were not associated with an enhanced risk of VTE and cardiovascular adverse events in a prospective study on patients with bladder cancer undergoing radical cystectomy.[28] However, the follow-up time after surgery was only 30 days; TE events occurring after this time point were not recorded in the study and might have been missed. Importantly, both this study and our analysis involved patients without apparent distant metastases. Hence, the total number of CTCs in localized disease states (including GBM) might be too low to induce a coagulopathy.
In support of this, in our study, the very low numbers of CTCs, ranging from 1 to 22 within 2.10 × 106 mononuclear cells,[9] is indicative that—if at all—higher absolute numbers of CTCs might be needed to account for an elevated TE risk in cancer.
Given that GBM surgery is a crucial risk factor for TE, it is therefore intriguing to determine whether patients with thromboembolic complications prior to GBM resection also exhibited CTCs beforehand. However, CTCs were not detected in all peripheral blood samples of the two patients who developed VTE before surgery. In our study, patients were not investigated for asymptomatic TEs which would have also been of interest when analyzing a potential thromboembolic effector role for CTCs in GBM.
A limitation of our study is the retrospective design. The TE incidence of 11% was lower than that in other documented GBM cohorts.[15]
[29] While our follow-up time ranges up to 128 months, the median follow-up time is 7.7 months. Because there was no difference in median follow-up times between patients with and those without TE, it seems unlikely that the results are biased by differential observation times. Since many patients were treated in ambulatory health care facilities following initial diagnosis and treatment, we cannot exclude that TE events occurring in the outpatient setting were not communicated or were missed.
To our knowledge, this is the first study investigating an association of CTCs and TE in GBM. While our study confirms a high risk of TE, CTCs were not a contributing factor to the incidence of TE in our patient cohort. Large prospective trials are needed to further investigate the effect of CTCs on thromboembolic complications in GBM.
