Keywords cancer-associated thrombosis - paraneoplastic coagulopathy - tissue factor - endometrial
cancer - microvesicle
Schlüsselwörter Krebs-assoziierte Thrombose - paraneoplastische Koagulopathie - Gewebefaktor - Endometriumkarzinom
- Mikrovesikel
Introduction
Cancer-associated thrombosis (CAT) is a frequent complication in patients with solid
tumors or hematological malignancies.[1 ] CAT significantly reduces quality of life and indicates an unfavorable clinical
outcome of the underlying malignancy.[1 ]
[2 ]
[3 ] Furthermore, CAT directly contributes to mortality in hematology and oncology patients.[3 ]
[4 ]
In general, the risk of venous thromboembolism (VTE) is increased four- to ninefold
in patients with cancer.[1 ]
[5 ] However, the risk is highly heterogeneous among individual cancer entities, with
highest VTE incidence rates reported in tumors of pancreatic, gastric, brain, or genitourinary
(excluding prostate) origin.[6 ] Moreover, in up to 11% of spontaneous VTE cases, an underlying malignancy is identified.[7 ] Finally, prevention and treatment of CAT are challenging since cancer patients are
at particularly high risk for both, recurrent thrombosis and bleeding, during anticoagulant
therapy.[3 ]
[8 ]
Here, we present the case of a 51-year-old woman who developed recurrent VTE and arterial
thromboembolism (ATE) despite standard therapeutic anticoagulation as the initial
manifestation of advanced endometrial cancer.
Case Presentation
A 51-year-old woman (175 cm, 70 kg) was transferred to our institution in August three
years ago for suspected paraneoplastic coagulopathy due to cancer of unknown primary
(CUP) with peritoneal and lymphatic metastases. The patient had no history of symptomatic
COVID-19. In addition, twice weekly PCR testing for SARS-CoV-2 remained negative during
hospitalization.
Three months earlier, the patient had developed deep vein thrombosis (DVT) of the
right lower extremity ([Fig. 1 ]). Anticoagulation with rivaroxaban was initiated, but despite reliably good adherence
the patient experienced an acute episode of amaurosis fugax, abdominal pain, tachycardia,
and shortness of breath 4 weeks later requiring hospital admission. Computed tomography
scanning of the chest and abdomen revealed right-sided pulmonary embolism (PE), embolic
infarction of the right upper kidney, and a suspicious lymph node located in perisigmoid
fat tissue with no evidence for a primary cancer. In addition, a thrombus in the right
atrium was revealed by echocardiography. There was no sign for an atrial shunt or
atrial fibrillation. Investigations for primary tumor site, including gastroscopy,
colonoscopy, and gynecological examination, remained unremarkable. Thus, anticoagulation
was continued with the vitamin K antagonist, phenprocoumon, at dosages titrated to
maintain the international normalized ratio between 3 and 4, and the patient was discharged
home from hospital. Several days later, painful swelling of the left leg occurred,
and DVT of the upper and lower extremity was confirmed by ultrasound ([Fig. 1 ]). Moreover, progressive PE and cerebral infarctions in addition to new thrombotic
deposits on mitral valve leaflets were observed. Anticoagulation was continued with
fondaparinux at 7.5 mg once daily (OD), and extensive laboratory workup ruled out
antithrombin deficiency, overt disseminated intravascular coagulation, antiphospholipid
syndrome, paroxysmal nocturnal hemoglobinuria, and autoimmune (or spontaneous) heparin-induced
thrombocytopenia ([Table 1 ]). In addition, molecular testing for JAK2V617F , prothrombin G20210A, and factor V gene mutation Leiden was negative. Tumor markers
CA125, CA15–3, CA19–9, and CEA were elevated. Follow-up computed tomography and magnetic
resonance imaging studies of abdomen and pelvis revealed progressive tumor manifestations
with multiple pelvic lymphatic and peritoneal lesions. Moreover, suspicious endometrial
alterations were suggestive of a gynecological cancer. Because of clinically progressive
coagulopathy, anticoagulation with fondaparinux was intensified to 7.5 mg three times
daily, and the patient was transferred to our institution for further diagnostic workup.
Table 1
Laboratory workup of the patient in July
Value
Reference range
Blood counts
Hemoglobin, g/dL
12.4
14.0–17.5
Leukocytes, 109 /L
10.5
3.8–11.0
Platelets, 109 /L
159
150–350
Clinical chemistry
Creatinine, mg/dL
0.5
0.6–1.3
LDH, U/L
292
87–241
Coagulation parameters
Prothrombin time, %
88
80–130
INR
1.0
0.85–1.15
aPTT, s
21
25–38
Thrombin time, s
16
16–22
Fibrinogen, g/L
3.7
1.8–4.0
D-dimer, mg/L
> 20
< 0.5
Antithrombin, %
103
83–118
PC activity, %
59
70–140
Free PS antigen, %
52
60–114
Ratio APC resistance
1.1
> 0.7
Factor VIII:C, %
129
70–150
Factor XIII:C, %
123
70–140
Plasminogen, %
110
75–140
Plasmin inhibitor, %
117
80–120
Tumor markers
CA125, kU/L
1,378
< 35
CA15–3, kU/L
92.4
< 25
CA19–9, kU/L
4,271
< 37
CEA, µg/L
47.3
< 3.8
Autoantibodies
IgM–aCL, U/mL
2.3
< 10
IgG–aCL, U/mL
2.7
< 10
IgM–anti-β2 GPI, U/mL
< 0.9
< 7
IgG–anti-β2 GPI, U/mL
0.8
< 7
LA
Negative
Negative
Genetic analysis
JAK2, pV617F
Negative
Negative
F5 mutation, Leiden
Negative
Negative
Prothrombin G20210A mutation
Negative
Negative
Abbreviations: aCL, anticardiolipin; anti-β2 GPI, anti-β2 -glycoprotein-I; APC, activated protein C; aPTT, activated partial thromboplastin
time; CA, cancer antigen; CEA, carcinoembryonic antigen; F5 , gene for coagulation factor V; INR, international normalized ratio; JAK2, Janus
kinase 2; LDH, lactate dehydrogenase; LA, lupus anticoagulant; PC, protein C; PS,
protein S.
Notes: EDTA-anticoagulated whole blood was used for genetic analysis and blood counts.
Clinical chemistry parameters were measured in lithium heparin-anticoagulated plasma
and coagulation parameters were measured in platelet-poor plasma obtained by centrifugation
of sodium citrate-anticoagulated whole blood for 15 minutes at 2,755 × g and 15 °C. Serum was used to quantify tumor markers and antiphospholipid antibodies.
Fig. 1 Clinical course of the patient. ASA, acetylsalicylic acid; AT, arterial thrombosis;
CT, cardiac thrombosis; DVT, deep vein thrombosis; PE, pulmonary embolism; VKA, vitamin
K antagonist. Asterisk indicates time point of laboratory workup ([Table 1 ]).
Anticoagulation was switched to the direct thrombin inhibitor, argatroban ([Fig. 1 ]). Compared with healthy controls, significantly elevated concentrations of microvesicle-associated
tissue factor procoagulant activity (MV TF PCA) were detected in patient plasma ([Fig. 2A ]), as assessed by a previously described chromogenic Xa generation endpoint assay.[9 ] High dosages of argatroban were required to maintain the activated partial thromboplastin
time within the upper therapeutic target range ([Fig. 2B ]). With the exception of an additional transient ischemic attack at the beginning
of argatroban therapy, which prompted initiation of antiplatelet therapy with acetylsalicyclic
acid (ASA) 100 mg OD, no further thrombotic events occurred ([Fig. 1 ]). Positron emission tomography scanning confirmed a tumorous process located in
the uterus with cervical, rectal, and bowel infiltrations and multiple lymphatic and
peritoneal metastases. While initial endometrial and cervical biopsy specimens were
nondiagnostic, a percutaneous core biopsy of a peritoneal tumor manifestation revealed
a highly estrogen and progesterone receptor positive endometrial adenocarcinoma, thus
confirming the paraneoplastic etiopathogenesis of coagulopathy.
Fig. 2 Characterization of paraneoplastic coagulation disorder. (A ) Tissue factor–dependent procoagulant activity of plasma microvesicles (MV TF PCA)
was assessed by a chromogenic factor Xa generation assay. MVs were isolated from patient
plasma on four distinct occasions during early anticoagulation with argatroban. Plasma
MVs obtained from three sex-matched healthy individuals served as controls. For MV
isolation, sodium citrate-anticoagulated whole blood was centrifuged within 1 hour
after collection twice at 2,050 × g and 4 °C for 10 minutes to obtain platelet-poor plasma (PPP). Subsequently, MVs were
isolated from PPP by double high-speed centrifugation at 16,100 × g and 4 °C for 30 minutes. (B ) Time course of activated partial thromboplastin time (aPTT) and argatroban dosages.
Target range is defined as 1.5- to 3.0-fold prolongation of the patient's baseline
aPTT and highlighted by horizontal dashed lines. (C ) Time course of tumor markers CA125 and CA19–9. Application of each cycle of neoadjuvant
chemotherapy (CTx) and upper limit of the reference range of tumor markers (dashed
lines) are indicated. Time course of patient MV TF PCA (D ) and D-dimer levels (E ) are shown. In panel (E ), the upper limit of the D-dimer reference range is indicated by a dotted line. Correlations
of D-dimer with MV TF PCA (F ) and tumor markers CA125 (G ) and CA19–9 (H ). All data were normally distributed using the Shapiro–Wilk test. p -Values are according to two-sample t -test (***, p < 0.001) in panel (A ). Correlation coefficients (r ) and p -values are according to the method of Pearson in panels F–H .
The endometrial cancer was classified as stage FIGO IVB, and the patient received
neoadjuvant chemotherapy with carboplatin and paclitaxel. Inhibition of tumor angiogenesis
with the monoclonal vascular endothelial growth factor antibody, bevacizumab, was
waived because of recurrent thromboembolic events. The patient showed a good clinical
response with decreasing tumor markers ([Fig. 2C ]). Consistently, release of MV TF PCA ([Fig. 2D ]) and D-dimer levels ([Fig. 2E ]) steadily declined during chemotherapy. In addition, MV TF PCA ([Fig. 2F ]), CA125 ([Fig. 2G ]), and CA19–9 ([Fig. 2H ]) strongly correlated with plasma D-dimers. Since tumor response was accompanied
by an improvement in coagulopathy, anticoagulation was continued with the low-molecular-weight
heparin (LMWH), enoxaparin, at supratherapeutic dosages (1.25 mg/kg twice daily [BID]).
After completion of all six chemotherapy cycles, the enoxaparin dosage was reduced
to 1 mg/kg BID. Cancer therapy was continued with debulking surgery, resulting in
complete resection of gross tumor lesions, followed by adjuvant percutaneous and local
radiotherapy. Immunohistochemical staining of the resected mass confirmed endometrial
cancer with tumor cells showing abundant TF expression ([Fig. 3 ]). At 15 months of follow-up, the patient is still in complete remission. No further
thrombotic events have occurred.
Fig. 3 Histopathological analysis of resected tumor. (A, B ) Histopathological analysis of tissue specimens obtained from debulking surgery revealed
an endometrioid adenocarcinoma. (C, D ) Immunohistochemical staining of tumor cells for TF using a specific rabbit monoclonal
IgG antibody (clone, EPR22548–240; Abcam, Cambridge, UK). Images were captured with
×2 magnification in panels (A ) and (C ), and ×10 magnification in panels (B ) and (D ).
Discussion
In the patient discussed in this study, recurrent thromboembolism was triggered by
advanced endometrial cancer. Endometrial cancer is the 7th leading type of cancer
in women worldwide with even higher prevalence in developed countries of the western
hemisphere.[10 ] In the majority of cases, endometrial cancer is diagnosed at localized stages, since
abnormal uterine bleeding typically occurs early in affected women.[10 ]
[11 ] Bleeding, however, was absent in our patient, and thrombotic coagulopathy was the
first manifestation of the hitherto occult malignancy. VTE rates of up to 11.5% have
been reported in endometrial cancer, with highest risks observed in patients with
advanced tumors.[12 ]
[13 ]
[14 ]
[15 ] In addition, endometrial cancer was the third leading tumor entity identified in
women with VTE and occult cancer.[16 ] Also, VTE occurrence is associated with an adverse outcome in this patient population.[13 ]
The pathogenesis of cancer-associated coagulopathy is complex and involves a plethora
of cellular and molecular mechanisms. These include aberrant expression and activation
of coagulation factors, release of procoagulant MVs, and activation of platelets and
leukocytes (i.e., monocytes and neutrophils) in the proinflammatory tumor microenvironment.[6 ]
[17 ]
[18 ] In particular, aberrant expression of TF, the physiologic initiator of the extrinsic
coagulation protease cascade, on cancer cells and release of TF-bearing MVs have been
associated with highly prothrombotic tumor entities and overt tumor cell-induced coagulation
activation.[17 ]
[18 ]
[19 ] In addition, some tumor entities may also produce other coagulation factors, including
coagulation factor VII (FVII) or cancer procoagulant,[20 ]
[21 ] or initiate coagulation by activation of the contact-dependent intrinsic pathway.[22 ] In our patient, we identified strong tumor cell TF expression in tissue sections
obtained from debulking surgery ([Fig. 3C, D ]), a finding consistent with previous (pre-)clinical studies.[23 ]
[24 ] Moreover, dramatically increased levels of MV TF PCA, presumably released by TF-expressing
tumor cells, were measured in patient plasma at initial presentation to our institution,
which steadily decreased during effective anticancer therapy ([Fig. 2A, D ]) and strongly correlated with plasma D-dimers ([Fig. 2F ]). Despite a limited number of data points, tumor markers CA125 and CA19–9 showed
a strikingly close correlation with plasma D-dimer levels ([Fig. 2G, H ]). It is thus tempting to speculate that TF expressed by tumor cells and cancer cell-derived
MVs was the main driver of paraneoplastic coagulation activation. Similarly, high
dosages of argatroban were required to control the hypercoagulable state ([Fig. 2B ]), which is in line with significant TF-driven thrombin generation.
Interestingly, MV TF PCA had markedly decreased during anticoagulation with argatroban
([Fig. 2D ]), suggesting that thrombin contributed to CAT pathogenesis by mechanisms additional
to fibrin formation. Several previous (pre-)clinical studies have indicated that thrombin
participates in primary tumor growth, cell proliferation and survival, metastasis,
angiogenesis, inflammation, and oncogenesis, particularly via cleavage activation
of protease-activated receptors (PARs).[25 ]
[26 ]
[27 ]
[28 ] In the proinflammatory tumor microenvironment, thrombin also promotes leukocyte
trafficking and activation of platelets and endothelial cells.[25 ]
[26 ]
[27 ] Moreover, thrombin inhibitors such as dabigatran or ximelagatran may reduce cancer
growth and metastasis in mice, but effects are highly variable between individual
tumor models and direct oral anticoagulant (DOAC) treatment schedules.[29 ] Similarly, rivaroxaban prevented FXa-mediated PAR2 activation on murine tumor-associated
monocytes/macrophages, which has been implicated in immune evasion of cancer cells.[30 ] Thus, it is tempting to speculate that thrombin-directed anticoagulants may inhibit
cancer progression, thereby enhancing their antithrombotic potency in CAT therapy.
However, convincing data from prospective clinical trials are currently not available.[26 ]
Despite significant insights into the molecular basis of paraneoplastic coagulopathies
over the past decades, prevention and treatment of CAT remain challenging. Consistent
with our case report ([Fig. 1 ]), cancer patients are at increased risk of recurrent thrombosis, with risks correlating
with tumor stage and entity.[1 ]
[6 ] Thus, extended anticoagulation beyond 6 months is recommended, especially in patients
whose cancer is still active and who carry a high thromboembolic risk.[31 ]
[32 ]
[33 ]
[34 ]
[35 ] In this population, current treatment guidelines recommend anticoagulation with
LMWH or direct oral FXa inhibitors (i.e., apixaban, edoxaban, and rivaroxaban).[31 ]
[32 ]
[33 ]
[34 ] In addition, effective treatment of the underlying malignancy is of utmost importance
as it addresses the primary cause of the coagulopathy.
Another challenge is VTE recurrence despite therapeutic anticoagulation. Following
exclusion of treatment nonadherence, switching the anticoagulant class (e.g., from
LMWH to DOAC or vice versa) can be considered.[32 ] Alternatively, the dose of LMWH may be increased by 20 to 25%.[32 ]
In addition to VTE, cancer patients are at increased risk of ATE, including myocardial
infarction, stroke, and peripheral arterial occlusions.[36 ]
[37 ] Similar to VTE, the highest risks were observed in patients with brain, stomach,
or pancreatic cancer.[36 ]
Our patient, however, developed recurrent ATE and VTE despite therapeutic anticoagulation
with various agents ([Fig. 1 ]). While arterial embolic events were most probably caused by cardiac thrombotic
manifestations, TF-driven coagulation activation by tumor cells and released MVs was
likely the main pathway leading to venous thrombosis ([Fig. 2A, D, F ]). Only continuous anticoagulation with argatroban in combination with antiplatelet
therapy with ASA was sufficient to attenuate initial coagulation activation, while
effective anticancer treatment was required for long-term control of the coagulopathy.
In conclusion, in patients with catastrophic paraneoplastic coagulation activation,
synergistic effects of therapeutic anticoagulation and effective anticancer therapy
may be required to control the thrombotic storm.
