Gastrointestinal (GI) cancer is estimated to occur in more than 333,000 Americans
and resulted in 167,790 cancer-related death in 2020.[1] GI cancer encompasses a wide range of pathologies, including esophageal, gastric,
pancreatic, neuroendocrine, hepatobiliary, colorectal, and anal cancers.[2]
[3]
[4]
[5]
[6]
[7] The estimated death associated with pancreatic, hepatobiliary, and colon cancers
ranks among the top five of all malignancies. Additionally, GI cancer and its treatment
are associated with multiple morbidities, including bleeding, infection, and thromboembolic
diseases. Among those, venous thromboembolism (VTE) is a common complication that
is six times more prevalent in the GI cancer population compared with the general
population.[8] In addition, VTE can result in recurrent deep venous thrombosis (DVT), pulmonary
embolism (PE), and postthrombotic syndrome (PTS), which can lead to increased short-term
mortality and long-term morbidity. Therefore, it is important to understand the prevalence
and management of thromboembolic disease in GI cancer patients. In this article, we
will discuss the clinical significance and management options of GI cancer-related
thrombotic disease.
Clinical Significance and Pathophysiology
Clinical Significance
Cancer is associated with an increased risk of VTE. In comparison to the general public,
the incidence of VTE in cancer patients is markedly higher. In a recent matched cancer
versus noncancer patient cohort study, cancer patients had a hazard ratio of 4.7 for
the occurrence of VTE and an incidence rate of 13.9 cases per 1,000 patient-years.[9] With regard to GI cancer, the VTE incidences for esophageal, gastric, colon, and
liver cancers were 12.5, 15.4, 13.4, and 7.2 events per 1,000 patients, respectively.
Pancreatic cancer had a particularly high VTE incidence of 22.7 events per 1,000 patients.[10]
Clinically, VTE can manifest as DVT and PE. In a Dutch registry, 63.6% of the VTE
patients presented with DVT, whereas 32.4% of the patients suffered from PE.[10] Clinically, DVT patients can present with pain, swelling, and warmth in the affected
limb. Of course, DVT can embolize distal organs, particularly the pulmonary vasculature.
PE can be subdivided into massive, submassive, and low-risk based on the likelihood
of mortality. Low-risk PE patients may remain asymptomatic. However, patients often
present with dyspnea, pleuritic chest pain, cough, and hemoptysis. The feared sequelae
is hemodynamically significant PE that can lead to cardiopulmonary compromise and
death.[11]
Apart from the immediate morbidity, VTE can also lead to long-term complications such
as recurrent VTE and PTS. In the Dutch registry, 12.6% of the patients had recurrent
VTE episodes.[10] In comparison to the noncancer population, the rate of recurrent VTE is two- to
threefold higher in the cancer population.[12] Moreover, cancer patients with VTE have a two- to threefold increase in major bleeding
events in comparison to the noncancer VTE patients.[13] Perhaps, part of the reason is that malignant cells directly contribute to the pathogenesis
of recurrent VTE. On the other hand, VTE treatment (anticoagulation) and frequent
thrombocytopenia in cancer patients likely result in an increased rate of major bleeding.
The increased recurrent VTE and major bleeding events have led to an increase in cancer
patient mortality.[14] Indeed, VTE is a leading cause of death in cancer patients receiving chemotherapy.[15] Furthermore, cancer patients with VTE have markedly higher short-term and long-term
mortality rates than those without.[16]
Recurrent VTE is a major risk factor for PTS.[17] PTS affects 20 to 50% of the DVT patients within 1 to 2 years of the index DVT episode.[18] Up to 10% of the patients will develop severe PTS.[19] Clinically, most PTS patients present with leg pain, heaviness, varicose veins,
and/or swelling, with a minority progressing to experience skin changes and/or venous
ulcers.[20] There is no gold standard test to diagnose PTS, but diagnosis and assessment of
clinical severity can be aided by several scoring systems. The Clinical-Etiology-Anatomy-Pathophysiology
(CEAP) classification is useful in characterizing the chronic venous disease. On the
other hand, the PTS severity is often measured using the Villalta score, which is
regarded as the international standard for the diagnosis and stratification of PTS.[20] Most importantly, PTS affects cancer patients' quality of life (QOL). In a cohort
study, PTS patients had significantly worse disease-specific QOL scores than those
without. In addition, patients with severe PTS had more significant decrease in disease-specific
QOL measures than those with the milder form of PTS.[21] Persistent leg pain and swelling can prohibit cancer patients from performing basic
daily tasks such as walking and standing and can lead to significant psychological
burden.
Pathophysiology
The pathophysiology of GI cancer-associated VTE is complex. A thorough review of the
topic is beyond the scope of this article. However, Virchow's triad dictates that
the causes are broadly related to the prothrombotic state, venous stasis, and endothelial
injury. GI malignancy induces a prothrombotic state by increasing tissue factor (TF)
expression.[22]
[23] TF is a glycoprotein that binds to factor VII when activated. The TF–factor VII
complex activates factor X, which propagates a coagulation cascade termed the “extrinsic
pathway.” In addition, chemotherapy can induce tumor lysis, which releases prothrombotic,
intracellular components. For example, the chemotherapeutic agent cisplatin, which
is commonly used to treat colorectal and pancreatic cancer, is associated with an
increased level of von Willebrand factor.[24] Furthermore, chemotherapy and oncological surgery can induce direct endothelial
damage. In addition, surgical oncology patients are often immobilized, which results
in venous stasis.
The pathophysiology of PTS is not completely understood. The normal leg venous return
is determined by the leg muscle pump and unidirectional venous valves. The cause of
PTS is likely a combination of venous valve damage as a sequel of DVT, outflow obstruction,
endothelial inflammation, and other factors. There is considerable debate on whether
venous reflux or proximal venous occlusion plays a larger role in PTS development.[25]
[26] The final common pathway appears to be persistent venous hypertension leading to
edema, pain, and ulceration.[27]
Management
Anticoagulation
Anticoagulation remains the first-line therapy for GI cancer-associated VTE. Per the
current standard of care, anticoagulation with low-molecular-weight heparin (LMWH)
should be administered for 3 to 6 months after the index VTE incident. The safety
and effectiveness of LMWH in cancer patients have been established in several randomized
controlled trials. The CLOT trial, published in 2003, studied cancer patients with
VTE treated with either LMWH (dalteparin) or coumarin derivative for 6 months. The
authors have found that the dalteparin group had a significantly lower recurrent VTE
rate (8 vs. 16%) than the coumarin derivative group with similar bleeding rates (6
vs. 4%).[28] In the CATCH trial, patients with active cancer were treated with either LMWH (tinzaparin)
or warfarin for 6 months. The authors found similar recurrent VTE rates between the
two treatment groups (7.2 vs. 10.5%; p = 0.07), whereas the warfarin group had a significantly higher rate of nonmajor bleeding
(11 vs. 15%; p = 0.004).[29] More recently published trials, such as the DALTECAN and TiCAN trials, have shown
that extended LMWH treatment (6–12 months) was generally safe.[30]
[31]
Although LMWH has been shown to be superior to vitamin K antagonists, patients often
find self-injection cumbersome. The newer direct oral anticoagulation (DOAC) agents
negate the inconvenience of LMWH. Published in 2018, the Hokusai VTE Cancer Thrombosis
trial investigated the effectiveness of edoxaban versus dalteparin in treating cancer-associated
VTE for at least 6 months and up to 12 months. The results have shown that edoxaban
was noninferior to dalteparin (p = 0.006). Edoxaban was associated with a statistically nonsignificant decrease in
recurrent VTE rate (hazard ratio: 0.71; p = 0.09) and a statistically significant increase in major bleeding (hazard ratio:
1.77; p = 0.04). Furthermore, with regard to GI cancer, the edoxaban group had a higher rate
of major bleeding compared with those treated with dalteparin (13.2 vs. 2.4%).[32] Select-D is an ongoing trial randomizing patient to either rivaroxaban or dalteparin
treatment for a total of 6 months. The first phase results have shown that the cumulative
recurrent VTE rate was 11% for the dalteparin group versus 4% for the rivaroxaban
group at 6 months. In addition, the major bleeding rate was 4% for dalteparin and
6% for rivaroxaban.[33]
Despite its safety profile and efficacy, anticoagulation is associated with an inherent
risk of bleeding and recurrent VTE. In the secondary analysis of the CATCH trial,
there was a 15.3% incidence rate of clinically relevant bleeding in patients treated
with either LMWH or warfarin over a 6-month period.[34] In comparison to LMWH, DOACs have shown a similar rate of bleeding in a large, retrospective
analysis (13 vs. 11%; p = 0.746).[35] The anticoagulation management of GI-cancer associated VTE is even more challenging
when patients are thrombocytopenic, which is relatively common when patients are treated
with chemotherapy.[36] Patients with thrombocytopenia are more prone to bleeding, and the anticoagulation
regimen needs to be dose-adjusted and closely monitored.[37] The International Society on Thrombosis and Haemostasis recommended holding anticoagulation
if the platelet count is less than 25 × 109 L–1.[37] Furthermore, with any medical therapy, medication compliance remains an issue, especially
for patients on LMWH. Evidence has shown that more patients had to be switched from
LMWH to warfarin, possibly due to the concern for self-injection.[38] The newer generation of DOAC negates many of the drawbacks of warfarin and LMWH.
However, its efficacy is still limited by patient compliance.
Apart from bleeding, recurrent VTE on anticoagulation is not uncommon. In the CLOT
trial, 9% of the patients treated with LMWH and 17% of the patients treated with warfarin
suffered from recurrent VTE.[28] In the CATCH trial, 7.2% of the patients in the LMWH treatment group and 10.5% of
the warfarin-treated patients had recurrent VTE.[29] Therefore, in cases where patients have contraindication to anticoagulation or have
failed anticoagulation, an alternative form of thromboprophylaxis is required.
Inferior Vena Cava Filter for GI Cancer-Associated Thrombosis
Strong indications for inferior vena cava (IVC) filter placement include patients
with symptomatic PE or proximal DVT and active bleeding or major contraindication to anticoagulation (e.g., recent surgery,
intracranial metastasis, or severe thrombocytopenia). IVC filters may also be used
when there is a major documented failure of anticoagulation therapy, although in some
patients altering the anticoagulation regimen may be an sufficient.[39] Currently, the American Society of Clinical Oncology recommends IVC filter placement
in conjunction with anticoagulation if there is DVT recurrence and progression despite
optimal anticoagulation therapy. However, the recommendation is based on expert opinion.[40]
With the widespread introduction of retrievable IVC filters, filter placement is sometimes
perceived as a low-risk procedure that can protect patients from recurrent PE. Since
their advent, the number of IVC filters placed in cancer patients has expanded significantly.
One study estimated that 19.2% of all cancer patients received an IVC filter during
the course of the treatment. However, only 7.7% of the filters were placed in patients
with an absolute contraindication to anticoagulation.[41] Furthermore, it is important to realize that most retrievable filters were left
in place, which can lead to multiple long-term complications such as filter fracture,
migration, and perforation. The complication rates range anywhere between 2 and 20%.[39] Therefore, it is important to coordinate patient care between the interventional
team and the ordering service to ensure proper filter removal once the patient can
be anticoagulated.[42]
The primary clinical utility of the IVC filter is to prevent fatal PE. However, there
exists a paucity of high-quality data examining the safety, efficacy, and mortality
benefit of IVC filter in GI cancer patients. The immediate periplacement complication
rate is very low.[43] Further evidence has shown that IVC filter complications (filter thrombosis, migration,
perforation) in cancer patients are not significantly different than that of the general
public.[44]
With regard to filter efficacy, there are two randomized controlled trials of IVC
filters in VTE patients. The PREPIC1 trial enrolled 400 patients and randomized them
into either permanent filter placement with anticoagulation or anticoagulation alone.
The trial has shown a decrease in recurrent PE rates at 12 days in the filter group.
However, the benefit was counterbalanced with an increase in recurrent DVT rates at
2 years.[45] At 8-year follow-up, the PE protective effect persisted, whereas there was an absolute
increase in recurrent DVT risk.[46] The PREPIC2 study enrolled 400 patients and randomized them into either retrievable
IVC filter placement plus anticoagulation or anticoagulation alone. The authors have
found that at 3 months, there was no difference in the rate of recurrent PE or mortality.
However, the filter plus anticoagulation group had a significantly higher rate of
recurrent DVT.[47] Both PREPIC1 and PREPIC2 studies included cancer patients. However, the sample size
of the cancer patients was not sufficient for stratified analysis.
Specifically, for cancer-associated VTE, Barginear et al conducted a small prospective
randomized trial comparing fondaparinux with and without IVC filter placement; 25%
of the patients had either colon or pancreatic cancer. The authors have found no survival
benefit between the two groups at 3 years.[48] Hence, for patients who can tolerate anticoagulation, the use of IVC filters is
not supported by quality randomized data.
There exists a significant amount of variability in patient outcome. Brunson et al
conducted a retrospective population-based cohort study involving 14,000 patients.
The presence of VTE, cancer, and IVC filter placement was identified using the ICD-9-CM
(International Classification of Diseases, 9th Edition, Clinical Modification) codes.
To minimize confounding variables, propensity scoring was used by applying a logistic
regression model. To correct the immortal time bias, IVC filter insertion was used
as a time-dependent covariate. The authors have found that IVC filter placement was
not associated with an improvement in 30-day mortality or adjusted 180-day recurrent
PE risk.[49] On the other hand, Stein et al conducted a large retrospective study involving 266,692
patients. Patients with cancer, IVF filter placement, and PE were identified using
the ICD-9-CM codes. The authors found that for patients aged > 60 years, filter placement
was associated with a significantly lower in-hospital all-cause mortality (7.4 vs.
11.2%; p < 0.0001). Furthermore, the filter group had a significantly lower 3-month all-cause
mortality (15.1 vs. 17.4%; p < 0.0001).[50] The major drawback of the study is the lack of propensity matching. Furthermore,
unstable patients and those who received thrombolytic therapy were excluded from the
final analysis.
Regardless, the decision to place an IVC filter in GI cancer patients must be individualized.
It is reasonable to place an IVC filter when anticoagulation is absolutely contraindicated
or has failed. However, the benefit of the IVC filter must be weighed against the
risk of complications.[51] Mansour et al have shown that patients with stage IV metastatic cancer and IVC filter
insertion had a median survival of 1.31 months.[52] Therefore, the benefit of IVC filter in this population may be marginal at best.
Cather-Directed Thrombolysis in GI Cancer-Associated VTE
For GI cancer-associated VTE, the standard anticoagulation therapy can prevent thrombi
extension but cannot dissolve the existing clot. In contrast, the additional administration
of fibrinolytic drugs actively dissolves thrombus, which may resolve venous obstruction
and improve the clinical status of the limb. The safety and effectiveness of catheter-directed
thrombolysis (CDT) in preventing PTS have been evaluated in a few randomized controlled
trials. The CaVenT trial is widely regarded as the first rigorous randomized controlled
trial. It enrolled 209 patients who were randomized to anticoagulation or anticoagulation
plus CDT in treating acute proximal DVT. Primary outcomes were assessed using the
Villalta score. The authors found no significant difference in the occurrence of PTS
at 6 months (30.3 vs. 32.2%; p = 0.77). However, there was a significant difference at 24 months (41.1 vs. 55.6%;
p = 0.047). In the CDT group, 3.3% of the patients experienced major bleeding.[53] At 5-year follow-up, CDT was shown to be persistently superior to anticoagulation
alone in preventing PTS (p < 0.0001), with an apparent increase in the size of the effect. However, at no time
point beyond 6 months was QOL improved by use of CDT.[54]
A drawback of the conventional CDT lies in its prolonged exposure to lytic agents,
where the treatment lasted 1 to 4 days in the CaVenT trial. Newer CDT techniques combine
both the lytic agent infusion and use of mechanical thrombectomy devices. Therefore,
the combination (pharmacomechanical CDT [PCDT]) decreases the patient exposure to
lytic agents and is therefore theoretically safer for patients.[55]
The large, NIH-sponsored ATTRACT (Acute Venous Thrombosis: Thrombus Removal with Adjunctive
Catheter-Directed Thrombolysis) trial enrolled 692 patients with proximal DVT and
randomized them to either anticoagulation or anticoagulation plus PCDT. Patients were
followed for 24 months. Patients with active cancer were excluded from this study.
The authors have found that over 24 months, there was no significant difference in
PTS prevention. Furthermore, major bleeding occurred in 1.7% of the PCDT group versus
0.3% in the control group (p = 0.049).[56] On first glance, the ATTRACT trial may show that PCDT does not necessarily improve
patient outcome. However, it is important to note that the PTS severity was significantly
lower in the PCDT group from 6 to 24 months. Furthermore, on stratified analysis,
the ATTRACT trial has shown the likely efficacy of PCDT in selected patient groups.
For patients with femoropopliteal DVT, PCDT was not shown to improve PTS prevention,
PTS severity, or QOL measurements.[57] However, for patients with acute DVT involving the iliac or common femoral veins
(iliofemoral DVT), PCDT was shown to provide greater reduction in presenting leg pain
and swelling and to significantly decrease the PTS severity at 6 to 24 months.[58] A subsequent detailed analysis of QOL found that in acute iliofemoral DVT, the use
of PCDT resulted in improved VEINES-QOL score as early as 1 month post-PCDT. The differences
were substantial at 1 month (10 points; p < 0.0001) and 6 months (8.8 points; p < 0.0001). The differences were still significant but smaller at 18 and 24 months
(5.8 and 6.6, p = 0.0086 and 0.0067 in per-protocol analyses, respectively).[59] On the other hand, PCDT was not cost-effective; at best, it may provide intermediate-value
care for the subgroup of patients with iliofemoral DVT ($137,000 dollars per quality-adjusted
life-year).[60]
Finally, the Dutch CAVA Trial randomized 184 acute iliofemoral DVT patients to receive
anticoagulation with or without additional ultrasound-assisted CDT. Evidence has shown
that ultrasound can cause disaggregation of fibrin fibers and that ultrasound pressure
waves increase lytic agent penetration into the thrombus.[61] Active cancer patients were excluded from the study. The primary outcome was assessed
using the Villalta score. The authors found no significant difference in terms of
PTS prevention at 12 months post-CDT (odds ratio: 0.75; 95% confidence interval: 0.38–1.5).
Furthermore, no significant difference was observed in terms of venous clinical severity
score or QOL. On the other hand, major bleeding occurred in four patients in the ultrasound-assisted
CDT group, whereas none occurred in the standard therapy group.[62]
Taken together, these studies suggest that CDT and related techniques do not provide
a significant clinical benefit that could justify use as the routine, first-line therapy
for patients with DVT. However, in symptomatic patients with extensive thrombus (i.e.,
iliofemoral DVT), these procedures appear to provide better relief of presenting symptoms
and may improve long-term QOL.
Despite its long history of use for DVT, CDT has not been routinely performed in cancer
patients. The reason is multifold. For one, early evidence suggested that cancer patients
may not derive durable benefits from CDT. Bjarnason et al have shown that the 2-year
primary patency rate was 41% in patients with malignancy compared with 75% in patients
without malignancy after CDT treatment of iliofemoral DVT.[63] Second, CDT has been associated with an absolute increase in bleeding risk. A 2016
systemic review has shown that 9% of the CDT patients experienced a major bleeding
compared with 4% of the anticoagulation group. However, there was a nonsignificant
difference in the rate of intracranial bleeding in this systematic review.[64] Given that GI cancer patients are at an increased risk of bleeding due to the frequent
thrombocytopenia and the need for anticoagulation, the absolute risk increase due
to CDT needs to be taken into consideration. Furthermore, the Society of Interventional
Radiology quality improvement guidelines of thrombolysis state that intracranial metastasis
needs to be ruled out before CDT therapy, mostly due to the risk of fatal intracranial
bleeding.[65] Third, cancer patients with a short expected life span may not consider the risk
of PTS to be a major priority in their overall care.
There are no prospective randomized controlled trial data focusing on the GI cancer
population. However, a few retrospective studies have shown that CDT can be safe in
cancer patients. Kim et al performed 202 CDT in patients with acute iliofemoral or
brachiosubclavian DVT. They have found that the rate of major bleeding was 4.9% in
the cancer patients and 3.4% in the noncancer patients (p = 0.6924). It is noteworthy that 75% of the major bleeding events in the noncancer
cohort occurred at the access site, whereas the majority of bleeding events in the
cancer cohorts were GI bleeding.[66] Furthermore, Brailovsky et al conducted a retrospective observational study in 1,290
cancer patients with proximal lower extremity DVT or vena-caval DVT who were treated
with CDT. The authors used propensity scoring to minimize the effect of confounding
variables. There was no significant difference in in-hospital mortality rate (1.9
vs. 2.6%; p = 0.23) or GI bleeding rate (2.3 vs. 2.2%; p = 0.89). However, the CDT group had a significant increase in the rate of intracranial
hemorrhage (1.3 vs. 0.4%; p = 0.02).[67]
Although the risk of bleeding must be weighed carefully, CDT of iliofemoral DVT may
be a useful therapeutic option for GI cancer patients who have severe clinical manifestations
such as acute limb-threatening circulatory compromise or (more commonly) severe pain
and swelling that limits ambulation despite initial anticoagulation. Patient selection
remains critically important, and individual risk-and-benefit analysis must be performed
before proceeding with CDT. Patients with intracranial metastasis must be excluded
given the absolute increase in intracranial bleeding associated with both intracranial
metastasis and CDT. In the near future, it is hoped that technical advances in thrombectomy
devices may decrease the risk of CDT in GI cancer patients through reduced dose and
duration of thrombolytic drug infusion. Lastly, as seen with CDT in stroke management,
the improvement in institutional and technical expertise is likely to make CDT safer
and more effective in managing DVT and in improving the health of cancer patients
with DVT.
