Predictors of venous thromboembolism development before and during chemotherapy for advanced germ cell tumor

Abstract

Objective

We retrospectively analyzed the incidence and localization of venous thromboembolism in patients undergoing chemotherapy for advanced germ cell tumor and separately evaluated the risk factors for venous thromboembolism development before and during chemotherapy.

Methods

We included 121 patients treated with cisplatin-based chemotherapy between 2005 and 2018. Venous thromboembolism was defined as venous thrombosis diagnosed using radiological imaging with or without thromboembolic symptoms. We analyzed the clinical parameters for identifying the possible venous thromboembolism risk factors. Khorana score was used to calculate the venous thromboembolism risk.

Results

Thirteen patients showed prechemotherapy venous thromboembolism and 13 developed venous thromboembolism during chemotherapy. The most common venous thromboembolism was deep vein thrombosis (10 patients), followed by inferior vena cava thrombus (eight patients) and pulmonary thrombus (six patients). Compared to the group without venous thromboembolism, the group with prechemotherapy venous thromboembolism showed higher proportion of patients with tumors originating in the right testis (10 out of 13), significantly higher lactate dehydrogenase levels (828 IU/L versus 436 IU/L, P = 0.013), significantly higher proportion of patients with retroperitoneal lymph node (RPLN) metastases >5 cm in diameter (76.9% versus 33.7%, P = 0.003) and slightly higher proportion of patients with high-risk Khorana score (≥ 3; 30.8% versus 11.6%). No significant differences were observed between the clinical characteristics of patients with venous thromboembolism developed during chemotherapy and patients without venous thromboembolism.

Conclusions

We show that both RPLN mass > 5 cm and high lactate dehydrogenase levels are significant risk factors for prechemotherapy venous thromboembolism but not for venous thromboembolism development during chemotherapy.

germ cell tumor, venous thromboembolism, cisplatin

Introduction

Patients with advanced germ cell tumor (GCT) have a favourable prognosis when treated appropriately with cisplatin-based chemotherapy and surgery. However, the treatments for advanced GCT are accompanied with several comorbidities. Venous thromboembolism (VTE) is a potentially fatal complication associated with cisplatin-based chemotherapy. A large retrospective study revealed high incidence of thromboembolic events in patients with various types of cancers treated with cisplatin-based chemotherapy (1). The reported incidence was 18.1% (169 of 932 patients) with a range of 0–38.5%, depending on the cancer type. A risk evaluation study showed that the rate of thromboembolic events in patients undergoing chemotherapy for GCT (19%) was significantly higher than that in a matched control group of patients undergoing cisplatin-based chemotherapy for cancers other than GCT (6%) (2). Subsequently, several investigators focussed on VTE events as complications during GCT chemotherapy and reported incidences between 5 and 24% (3–9).

It is well known that VTE development is associated with the release of procoagulants or cytokines by cancer cells (10). However, information on VTE incidence in advanced GCT before initiation of chemotherapy is rather limited (3,6,7). This is because most studies focused on the contribution of chemotherapy to VTE development and excluded pretreatment VTEs from the analyses (7).

VTE risk estimation is essential for timely diagnosis, prevention of further complications and identification of patient subgroups that would benefit from thromboprophylaxis. Therefore, Khorana et al. proposed a predictive model for chemotherapy-associated VTE. The Khorana model involves simple baseline clinical and laboratory variables, including body mass index (BMI), prechemotherapy blood cell count levels and the cancer site (very high risk: stomach and pancreas; high risk: lung, lymphoma, gynecologic, bladder and GCT) (11). Nevertheless, this model is not specific for GCT chemotherapy, and several investigators proposed other predictive factors for VTE development during GCT chemotherapy. However, to the best of our knowledge, only one report describes the risk factors for VTE development before chemotherapy in patients with GCT (3).

Therefore, in this retrospective study, we investigated the incidence and localization of VTE and evaluated the risk factors for VTE development before and during chemotherapy in patients with advanced GCT. Furthermore, we evaluated the efficacy of Khorana model in patients with GCT.

Patients and methods

Patients and treatment

We retrospectively analyzed 121 of the 123 patients with advanced GCT treated with cisplatin-based chemotherapy at Tsukuba University Hospital (TUH) between 2005 and 2018; two patients were excluded owing to the lack of sufficient data. At TUH, first-line chemotherapy involves 3–4 courses of bleomycin, etoposide and cisplatin (BEP), according to the International Germ Cell Cancer Collaborative Group (IGCCCG) risk criteria (12). To avoid bleomycin-induced pulmonary toxicity, etoposide and cisplatin (EP) or etoposide, ifosfamide and cisplatin (VIP) regimen is used instead of BEP for patients aged >50 years (13). It is standard practice at our institution to use EP/VIP during the first or second course of treatment for patients with aggressive lung metastases or who are high risk for other pulmonary complications. In some patients, BEP was changed to EP/VIP during chemotherapy, when pulmonary toxicity and/or VTE was suspected. The most commonly used regimen for second-line chemotherapy was paclitaxel, ifosfamide and cisplatin (14,15).

Data collection

Data were collected from individual medical records for all patients. The primary tumour localization, histology, TNM staging classification, IGCCCG prognosis group, prechemotherapy tumour marker levels, size of retroperitoneal lymph node (RPLN) metastases, chemotherapy regimen, post-chemotherapy surgery and oncological outcomes were recorded. The Khorana risk model (Table 1) was used for evaluating VTE risk based on BMI and prechemotherapy blood cell counts.

Table 1

Khorana model

Patient characteristics	Risk score
Site of cancer
Very high risk (stomach, pancreas)	2
High risk (lung, lymphoma, gynaecologic, bladder, germ cell)	1
Prechemotherapy platelet count ≥350 × 109/l	1
Haemoglobin <10 g/l or use of red cell growth factors	1
Prechemotherapy leucocyte count >11 × 109/l	1
Body mass index ≥35 kg/m2	1

Diagnosis of VTE

VTE was defined as venous thrombosis, diagnosed using radiological imaging, with or without thromboembolic symptoms. Before and during chemotherapy, we routinely performed contrast-enhanced chest, abdominal and pelvic computed tomography (CT). If VTE was suspected on the basis of CT or symptom development, further radiological investigation was performed using Doppler ultrasound, magnetic resonance imaging or 18F-fluorodeoxyglucose (FDG)-positron emission tomography (PET)/CT as needed. Medical records were reviewed with special attention to reports on imaging studies to identify VTE. VTE before chemotherapy and VTE between the first day of chemotherapy and the final day of the last chemotherapy were recorded.

Statistical analyses

Categorical variables were compared between cohorts using the Pearson’s chi-squared test for two or more variables. Continuous variables were compared using the Mann–Whitney U test. Survival curves were constructed using the Kaplan–Meier method. The overall survival (OS) was compared using the log-rank test. The relative influence of different prognostic factors on OS was estimated using Cox’s proportional hazards model with multiple variables. The level of significance was set at P < 0.05 for all the analyses. All statistical analyses were performed using SPSS® 25.0 for Windows® (SPSS Inc., Chicago, IL, USA).

Results

We first investigated the VTE events in 121 patients who underwent chemotherapy for advanced GCT. VTE was observed in 26 of the 121 patients (21.5%) who received chemotherapy for advanced GCT (Table 2). Of these, 13 patients showed VTE during imaging studies performed before chemotherapy initiation, whereas 13 patients newly developed VTE during chemotherapy. VTE was diagnosed during first-line and second-line chemotherapy in 10 and 3 patients, respectively. The most common VTE form was deep vein thrombosis (DVT) (10 patients) in lower extremities. This was followed by inferior vena cava (IVC) thrombosis (eight patients) and pulmonary thrombosis (six patients). On the basis of the findings from contrast-enhanced CT or 18F-FDG-PET/CT, four patients were diagnosed with pure blood IVC thrombus and four with blood thrombus combined with tumour thrombus. Radiographically, all tumour thromboses were considered as derived from RPLN metastasis. Six of the eight IVC thrombi developed before chemotherapy. Of the six patients with pulmonary thrombus, two showed complications with IVC thrombus and DVT. All these six patients showed no respiratory symptoms; however, two patients showed VTE symptoms, one patient with superior sagittal sinus thrombosis suffered from headache, and one patient experienced pain in the lower extremities because of DVT.

Table 2

Characteristics of VTE

	Total	VTE before chemotherapy	VTE during chemotherapy
Number	26	13 (50.0%)	13 (50.0%)
Number of site of DVT
Single site	20 (76.9%)	11 (84.6%)	9 (69.2%)
Two or more sites	6 (23.1%)	2 (15.4%)	4 (30.8%)
Localization of VTE
Deep vein thrombus	10 (38.4%)	4 (30.8%)	6 (46.2%)
Pulmonary thrombus	6 (23.1%)	2 (15.4%)	4 (30.8%)
Inferior vena cava thrombus	8 (30.8%)	6 (46.2%)	2 (15.4%)
Renal vein thrombus	3 (11.5%)	1 (7.7%)	2 (15.4%)
Common iliac vein thrombus	4 (15.4%)	2 (15.4%)	2 (15.4%)
Superior sagittal sinus thrombosis	1 (3.8%)	0 (0%)	1 (7.7%)

VTE, venous thromboembolism.

P < 0.05 was the level of significance.

Next, we compared the clinical characteristics of patients with VTE and those without VTE (Table 3). The proportion of patients with tumours originating in the right testis was higher in patients with prechemotherapy VTE (10 of 13 patients) than that in patients without VTE. Patients with prechemotherapy VTE also showed significantly higher lactate dehydrogenase (LDH) levels (828 versus 436 IU/l, P = 0.013) and significantly higher proportion of RPLN metastases >5 cm in diameter (76.9 versus 33.7%, P = 0.003) than those in patients without VTE. The proportion of patients with a high-risk Khorana score (≥3) was higher for patients with prechemotherapy VTE than that for patients without VTE (30.8 versus 11.6%); however, the difference was not statistically significant. There was no difference in the BMI, clinical stage, histology and IGCCCG risk criteria between both the groups. Conversely, no significant difference was observed in the clinical characteristics of patients with newly developed VTE during chemotherapy and patients without VTE (Table 3).

Table 3

Characteristics of patients with and without VTE

	Patients without VTE (N = 95)	Patients with VTE before chemotherapy (N = 13)	P*	Patients with VTE during chemotherapy (N = 13)	P**
Age
Median	34	38	0.175	30	0.140
Range	17–70	23–79		25–42
>35	47 (49.5%)	9 (69.2%)	0.181	5 (38.5%)	0.456
Body mass index
Median	22.2	21.6	0.795	22.9	0.644
Range	14.7–44.0	18.5–35.3		20.3–28.7
Primary tumour	n (%)	n (%)		n (%)
Testis	90 (94.7%)	12 (92.3%)	0.006	11 (84.6%)	0.395
Right	37 (41.1%)	10 (83.3%)		6 (54.5%)
Left	53 (58.9%)	2 (16.7%)		5 (45.5%)
RPLN	2 (2.1%)	0 (0%)		0 (0%)
Mediastinum	3 (3.2%)	1 (7.7%)		2 (15.4%)
Type of primary site			0.803		0.141
Seminoma	25 (26.3%)	3 (23.1%)		1 (7.7%)
Non-seminoma	70 (73.7%)	10 (76.9%)		12 (82.3%)
IGCCCG criteria			0.979		0.276
Good	36 (37.9%)	5 (38.5%)		2 (15.4%)
Intermediate	20(21.1%)	3(23.0%)		4(30.8%)
Poor	39 (41.0%)	5 (38.5%)		7(53.8%)
Clinical stage at presentation (UICC)			0.226		0.579
I	0 (0%)	0 (0%)		0 (0%)
II	21 (22.1%)	1 (7.7%)		2 (15.4%)
III	74 (77.9%)	12 (92.3%)		11 (84.6%)
Prechemotherapy tumour markers
HCG, median (range) mIU/ml	96 (0.1–8230000)	229 (0.4–399000)	0.806	4442 (0.7–1371250)	0.105
AFP, median (range)ng/ml	11 (0.9–40376)	6 (0.1–22580)	0.974	10 (1.2–21198)	0.744
LDH, median (range)U/l	436 (130–5577)	828 (349–7255)	0.013	619 (138–2416)	0.081
Khorana risk score			0.174		0.616
1	51 (53.7%)	5 (38.4%)		4 (30.8%)
2	31 (32.6%)	4 (30.8%)		4 (30.8%)
≥3	11 (11.6%)	4 (30.8%)	0.066	2 (15.4%)	0.460
Unknown	2 (2.1%)	0 (0%)		3 (23.0%)
RPLN >5 cm	32 (33.7%)	10 (76.9%)	0.003	5 (38.5%)	0.734
Chemotherapy
Second-line or more	47 (49.5%)	4 (30.8%)	0.205	9 (56.3%)	0.181
Platelet count, median (range) × 109/l	27.1 (15.7–60.3)			31.0 (15.6–45.7)	0.324
eGFR, median (range) ml/min/1.73m2	93.8 (38.5–2856)			85.5 (46.2–201)	0.453
D-dimer, median (range) μg/ml	1.7 (0.5–46)			3.4 (0.7–14.7)	0.082

RPLN, retroperitoneal lymph node; IGCCCG, International Germ Cell Cancer Collaborative Group; UICC, Union for International Cancer Control; HCG, human chorionic gonadotropin; AFP, alpha-fetoprotein; LDH, lactate dehydrogenase; eGFR, estimated glomerular filtration rate.

^*Comparison between patients without VTE and patients with prechemotherapy VTE.

^**Comparison between patients without VTE and patients with VTE developed during chemotherapy.

Since there is a possibility that patients with newly developed VTE might have different risk factors, we additionally analyzed pretreatment platelet count, eGFR and D-dimer in those patients. Pretreatment platelet count and eGFR were available for most of patients. In contrast, data of D-dimer were missing from some patients’ records but were available in 9 of 13 patients with VTE during chemotherapy and 47 of 95 patients without VTE, respectively. As shown in Table 3, there was no statistical difference in platelet count, eGFR and D-dimer between patients with newly developed VTE during chemotherapy and patients without VTE (P = 0.324, 0.453, 0.082, respectively).

We further investigated the association of treatment procedures with VTE by comparing the treatment regimens used in patients with or without VTE (Table 4). When VTE was diagnosed, we routinely consulted to cardiologists for advice on anticoagulant therapy; the most commonly used agent was sodium heparin, followed by edoxaban and warfarin. In recent cases, however, direct oral anticoagulants, such as edoxaban, were more frequently selected than sodium heparin or warfarin. During first-line chemotherapy, the initially selected regimen was BEP in 15 (57.7%) and EP or VIP in 8 (30.7%) of the 26 patients with VTE; however, BEP was changed to VIP in 3 of these 15 patients. Thus, 12 of the 26 patients with VTE (46.2%) received BEP monotherapy as first-line chemotherapy. Similarly, 70 of the 95 patients without VTE (73.7%) initially received BEP, and BEP was changed to EP or VIP in 13 patients (13.7%) during the first-line chemotherapy. Thus, 57 of the 95 patients without VTE (60.0%) received BEP monotherapy. The proportion of patients receiving BEP monotherapy as first-line chemotherapy was lower in the group with VTE than that in the group without VTE; however, the difference was not statistically significant (P = 0.206). Dose reduction and postponing chemotherapy cycles over 3 days were not needed in patients with VTE. Therefore, the presence of VTE was not associated with a decrease in chemotherapy intensity.

Table 4

Treatment options for patients with and without VTE

	Patients with VTE (N = 26)	Patients without VTE (N = 95)	P
First-line chemotherapy
BEP monotherapy	12 (46.2%)	57 (60.0%)	0.206
VIP monotherapy	4 (15.4%)	8 (8.4%)
EP	2 (7.7%)	10 (10.5%)
BEP → VIP	3 (11.5%)	11 (11.6%)
VIP → BEP	2 (7.7%)	7 (7.4%)
BEP → EP	0 (0%)	2 (2.1%)
Others	3 (11.5%)	0 (0%)
Second-line chemotherapy	13(50.0%)	46 (48.5%)	0.887
TIP	13 (50.0%)	41 (43.2%)
VIP	0 (0%)	0 (0%)
Others	0 (0%)	5 (5.3%)
Post-chemotherapy surgerya	18(69.2%)	69(72.6%)	0.732
RPLND	11 (42.3%)	45 (47.4%)
Thoracotomy	6 (23.1%)	18 (18.9%)
Others	1 (3.8%)	6 (6.3%)
Anticoagulant therapya
Heparin sodium	13 (50.0%)	–
Warfarin	10 (38.5%)	–
Edoxaban	12 (46.2%)	–
Apixaban	2 (7.7%)	–
Dabigatran	1 (3.8%)	–
Rivaroxaban	1 (3.8%)	–

BEP, bleomycin, etoposide, and cisplatin; VIP, etoposide, ifosfamide, and cisplatin; EP, etoposide and cisplatin; RPLND, retroperitoneal lymph node dissection.

a

^aOverlapping therapies.

The proportion of patients receiving second-line chemotherapy or further treatment after second-line chemotherapy and the proportion of patients who underwent post-chemotherapy surgery did not differ between the group with VTE and the group without VTE.

During the median follow-up of 39.4 months in patients with VTE, one patient in the poor-prognosis group died because of disease progression. In contrast, no cancer-related deaths were observed in the intermediate- and good-prognosis groups. Thus, the 5-year OS rate in patients with VTE was 100% in the intermediate- and good-prognosis groups and 88.8% in the poor-prognosis group. The 5-year OS rates in patients without VTE were 100, 82.5 and 86.4% in the good-, intermediate- and poor-prognosis groups, respectively. There was no statistical difference between patients with VTE and patients without VTE in respective groups of IGCCC prognosis group (P = 0.266 and 0.913 in intermediate- and poor prognosis groups, respectively).

We performed multivariate Cox proportional analysis in OS after chemotherapy using parameters of IGCCCG prognosis group, with or without VTE, and BEP monotherapy or other regimens as induction chemotherapy. These parameters were not significant prognostic factors (P = 0.965, 0.642, 0.17, respectively).

Discussion

Cancer itself and the cisplatin-based chemotherapy administered are known to be risk factors for VTE development. Therefore, patients with advanced GCT treated using chemotherapy, where cisplatin is a key drug, are at a higher risk for VTE. Of the 121 patients included in the present study, VTEs were observed before chemotherapy in 13 patients (10.7%) and VTEs developed during chemotherapy in 13 patients (10.7%); the total VTE incidence was 21.5% (26 out of 121 patients). The incidence of VTE development during chemotherapy is consistent with the values reported previously (3–9). In contrast, incidence of prechemotherapy VTE is higher than that reported by previous studies (2.6–9.3%) (3,6,7). Therefore, it is important to recognize the increased potential of a VTE complication at any time throughout GCT chemotherapy.

DVT in lower extremities was commonly observed in both prechemotherapy VTE and VTE developed during chemotherapy. In contrast, six of the eight IVC thrombi developed before chemotherapy. Precise diagnostic imaging via contrast-CT or 18F-FDG-PET revealed pure blood IVC thrombus in four patients and blood thrombus combined with tumour thrombus in four patients. Of the six patients with pulmonary thrombi, two showed both IVC thrombus and DVT. A small retrospective study previously reported that tumoural IVC thrombus could be treated with chemotherapy without development of pulmonary embolism (PE) (16). However, our results suggest that it is important to first identify whether the IVC thrombus is tumoural or benign or a combination of both.

Risk estimation of VTE aids in its timely diagnosis, provides an indication of thromboprophylaxis and helps to prevent any further complications. Several investigators reported that high-risk Khorana score, elevated LDH levels, febrile neutropenia and large RPLN mass (>5 cm in diameter) are predictors of VTE development during GCT chemotherapy (3,4,8,9). However, the reported results are not completely consistent; in particular, Khorana score and febrile neutropenia were not reported as significant predictors in some studies (2,5). In the present study, we separately analyzed the risk factors for VTE development before and during chemotherapy in patients with advanced GCT. We found that tumours originating in the right testis, RPLN mass >5 cm and high LDH levels were significant risk factors for VTE development before chemotherapy, but not for VTE development during chemotherapy (Table 2). Fourteen of 16 patients with right testicular cancer with VTE had RPLN metastasis around the IVC. Therefore, we posit that physical pressure on the IVC is associated with development of VTE in those patients. Honecker et al. also separately analyzed the risk factors for prechemotherapy VTE and VTE development during chemotherapy and reported both elevated LDH levels and RPLN mass as predictors for prechemotherapy VTE but not for VTE development during chemotherapy (3). Patients with prechemotherapy VTE did not always show a high-risk Khorana score, compared to that in patients without VTE, and the score was not predictive for VTE development during chemotherapy. This discrepancy in observation may be partially attributed to the relatively small sample size of the present study. Furthermore, the exact onset of VTE was not reported in some of the previous studies (3).

Accurate diagnosis of VTE and the subsequent initiation of anticoagulant therapy are essential for avoiding fatal complications. Death due to PE was reported in 3 of 100 patients with GCT (3%) in a risk assessment study (2) and in 14 of 144 cases (9%) in an autopsy study (17). In addition to the development of fatal PE, VTE is known to be associated with worse oncological outcomes. Gonzalez-Billalabeitia et al. (7) showed that VTE is an independent adverse prognostic factor for both progression-free survival (PFS) and OS, especially for patients in the IGCCG intermediate-prognosis group. They also highlighted that the presence of VTE before chemotherapy was a worse prognostic factor for both PFS and OS. In contrast, in the present study, further complications associated with VTE were not observed, and only 1 of the 26 patients with VTE died owing to disease progression. Therefore, the adverse effect of VTE on oncological outcome remains poorly understood, and further investigation is warranted to confirm this.

Among patients receiving cisplatin-based chemotherapy, the thromboembolic event rate in patients with GCT was reported to be significantly higher than that in patients with cancers other than GCT (2). This finding suggests that cisplatin is not the only risk factor for VTE during GCT chemotherapy. Although the precise mechanism underlying the development of VTE during chemotherapy remains unknown, endothelial vascular damage and possible alterations in the clotting cascade are considered to be responsible for chemotherapy-related DVT (18). Several chemotherapeutic agents are known to cause vascular toxicity, and cisplatin has been reported to cause potent endothelial vascular damage via direct toxic effect or free radical generation. However, other drugs used in GCT chemotherapy, especially bleomycin, have been reported to cause vascular toxicity via a different mechanism (18). As shown in Table 3, we tended to use EP or VIP instead of BEP after diagnosis of VTE, especially when PE or VTE with high risk for PE is presented. This change in chemotherapy regimen is recommended to avoid further bleomycin-induced pulmonary damage and any additional risk of vascular injury. However, at present, there is no clear evidence to support avoiding bleomycin use for patients with VTE.

However, the present study has a few limitations. First, the study population is relatively small. Second, owing to the retrospective design, we could not rule out the possibility of selection bias. Nevertheless, our data show that VTE-related complications occur in 21.5% patients undergoing chemotherapy for GCT and that half of these patients showed VTE before chemotherapy initiation. We suggest that patients with prechemotherapy VTE should be assigned to high-risk groups and administered appropriate anticoagulant therapy for effective VTE management.

In summary, we have showed RPLN metastases >5 cm, and high LDH levels are significant risk factors for prechemotherapy VTE. Further investigations are needed to elucidate the risk factors for VTE development during chemotherapy.

Conflict of interest statement

None.