Venous Thromboembolism andNonsmall Cell Lung Cancer

A Pooled Analysis of National Cancer Institute of CanadaClinical Trials Group Trials

Lisa K. Hicks, MD, MSc1,2; Matthew C. Cheung, MD, MS2,3; Keyue Ding, PhD4;

Baktiar Hasan, PhD4; Lesley Seymour, MD, PhD4; Aurelie Le Maıtre, MSc4;

Natasha B. Leighl, MD, MSc2,5; and Frances A. Shepherd, MD2,5

BACKGROUND: Advanced nonsmall cell lung cancer (NSCLC) is associated with venous thromboembolism

(VTE). However, to the authors’ knowledge, the incidence of VTE in early NSCLC, predictors of VTE, and

the prognostic significance of VTE in NSCLC have not been explored. METHODS: Individual patient data

from 3 National Cancer Institute of Canada Clinical Trials Group trials were analyzed (n ¼ 1987 patients).

Clinical Trial BR.10 was a randomized study of postoperative vinorelbine and cisplatin versus observation in

patients with stage IB/II NSCLC (grading determined according to the TNM staging system). Clinical Trial

BR.18 was a randomized study of paclitaxel and carboplatin with or without the metalloproteinase inhibitor

BMS-275291 in patients with advanced NSCLC. BR.21 was a randomized study of erlotinib versus placebo in

patients with previously treated NSCLC. The relations between VTE, treatment, concomitant medications,

and patient characteristics were explored in univariate and multivariate analyses. Survival analysis was

completed using Cox regression. RESULTS: The incidence of VTE ranged from 0% in patients with early

stage NSCLC on the observation arm of BR.10 to 7.9% in patients with advanced NSCLC who received

chemotherapy (BR.18). Patients with early stage NSCLC who received chemotherapy (BR.10) and patients

with previously treated NSCLC who received erlotinib or placebo (BR.21) had a VTE incidence of �3%. Fac-

tors that were found to be predictive of VTE included previous VTE (BR.18; P ¼ .001) and obesity (BR.10; P

¼ .03). In patients with advanced NSCLC, VTE was associated with shorter survival (BR.18: hazard ratio

[HR], 1.61; 95% confidence interval [95% CI], 1.26-2.07 [P ¼ .0002]; BR.21: HR, 2.18; 95% CI, 1.57-3.04 [P <

.0001]). CONCLUSIONS: In patients with both early stage and advanced stage NSCLC, VTE occurred more

frequently in patients who received chemotherapy (but not erlotinib or BMS-275291). In patients with

advanced stage NSCLC, VTE was associated with obesity and a history of VTE. VTE was found to be prog-

nostic in patients with advanced stage NSCLC. Cancer 2009;115:5516–25. VC 2009 American Cancer

Society.

KEY WORDS: lung neoplasms, complications, drug therapy, mortality, venous thromboembolism,

epidemiology, etiology, adjuvant chemotherapy, antineoplastic combined chemotherapy protocols.

Received: January 24, 2009; Revised: March 12, 2009; Accepted: March 28, 2009

Published online August 26, 2009 in Wiley InterScience (www.interscience.wiley.com)

DOI: 10.1002/cncr.24596, www.interscience.wiley.com

Corresponding author: Lisa K. Hicks, MD, MSc, Keenan Research Centre, Li Ka Shing Knowledge Institute, St. Michael’s Hospital, 30 Bond Street,

Toronto, Ontario, Canada M5B 1X8; Fax: (416) 864-3073; hicksl@smh.toronto.on.ca

1Keenan Research Center, Li Ka Shing Knowledge Institute of St. Michael’s Hospital, Toronto, Ontario, Canada; 2Faculty of Medicine, University of

Toronto, Toronto, Ontario, Canada; 3Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; 4National Cancer Insti-

tute of Canada Clinical Trials Group, Kingston, Ontario, Canada; 5University Health Network, Princess Margaret Hospital, Toronto, Ontario, Canada

5516 Cancer December 1, 2009

Original Article

An association between cancer, including nonsmall celllung cancer (NSCLC), and venous thromboembolism

(VTE), has been demonstrated previously.1,2 Large regis-

try studies suggest that the relative risk of VTE in the set-

ting of malignancy is approximately 6.51 and that VTE is

more common in advanced cancer.1,2 The association

between VTE and malignancy is most likely multifacto-

rial, with tumor biology, disrupted homeostasis, the use

of central venous catheters, surgery, patient immobility,

concomitant medications, and systemic cancer therapy all

contributing.3,4

The correlation between cancer treatment and VTE

risk has been explored most thoroughly for breast cancer

and multiple myeloma. In breast cancer, both adjuvant

chemotherapy and hormone therapy increase the risk of

VTE.5-8 In myeloma, thalidomide increases the risk of

VTE, particularly when administered with chemotherapy

or dexamethasone.9,10 By using registry data, Chew et al

reported a 1-year cumulative incidence of VTE of 3% for

patients with all stages of NSCLC and small cell lung can-

cer.11 In their model, advanced stage was a significant pre-

dictor of VTE in patients with NSCLC.11 Khorana et al12

reported an odds ratio for VTE of 1.86 (95% confidence

interval [95% CI], 0.67-3.38) in ambulatory lung cancer

patients who were receiving chemotherapy. An increased

risk of VTE also was reported with the use of the metallo-

proteinase inhibitor prinomastat in patients with

advanced NSCLC.13 However, the issues of whether there

is a significant risk of VTE in early stage NSCLC, which

factors predict for VTE occurrence, whether chemother-

apy has an impact on the risk of VTE in NSCLC, and

whether VTE is associated with worse clinical outcomes

have been explored less thoroughly.

To address these issues, we undertook a retrospective

analysis of 3 randomized trials in NSCLC conducted by

the National Cancer Institute of Canada (NCIC) Clinical

Trials Group (CTG), 2 of which included a no-treatment

arm. Herein, we report the result of that analysis.

MATERIALS AND METHODS

Trials and Patients

Individual patient data from 3 NCIC CTG clinical trials

(BR.10, BR.18, and BR.21) were included in this retro-

spective cohort study. Complete study details for each

trial have been published previously.14-16 Briefly, BR.10

was a randomized study that compared 4 cycles of postop-

erative adjuvant vinorelbine and cisplatin with observa-

tion in patients with completely resected stage IB and II

NSCLC (grading determined according to the TNM

staging system).14 BR.18 was a randomized study of first-

line chemotherapy with paclitaxel andcarboplatin with or

without the metalloproteinase inhibitor BMS-275291 in

patients with incurable stage IIIB/IV NSCLC.15 BR.21

was a randomized study of erlotinib versus placebo (2:1

randomization) in patients with stage IIIB/IV NSCLC

who had received previous systemic chemotherapy.16

Statistical Analysis

All statistical analyses were performed using SAS software

(Statistical Analysis System, Version 9.1; SAS Institute Inc.,

Cary, NC). Analyses were stratified by treatment arm.

Results were considered significant with a 2-sided P value of

�5%. BR.18 and BR.21 were analyzed separately and as a

combined dataset, because these 2 trials evaluated patients

with advanced rather than early stage, resected NSCLC.

When appropriate, discrete variables were summar-

ized according to the number and proportion of patients

that fell into each category and were compared using a

chi-square test. Continuous and ordinal categoric varia-

bles were summarized using the mean, median, standard

error, and minimum and maximum values and, when

appropriate, were compared using the F test. VTE was

defined as: 1) a history of VTE, a history of pulmonary

embolus or deep venous thrombosis reported on baseline

case report forms; 2) VTE on study, grade 2, 3, 4 or 5

deep venous thrombosis or pulmonary embolus according

to version 2 of the Common Toxicity Criteria (Cancer

Therapy Evaluation Program, National Cancer Institute,

National Institutes of Health, Bethesda, Md) for BR.18

and BR.21; or, for BR.10, VTE was coded as CDVEN

(according to NCIC CTG Expanded Common Toxicity

Criteria) that occurred after randomization; and 3) VTE

at any time, a history of VTE, and VTE present at ran-

domization or on study.

Predictive Analyses of VTE on Study

Baseline variables that were evaluated for their association

with VTE on study included the following: treatment

VTE and NSCLC/Hicks et al

Cancer December 1, 2009 5517

allocation, age (�65 years or >65 years), sex, pathologic

subtype of NSCLC (adenocarcinoma, squamous, or

other), Eastern Cooperative Oncology Group perform-

ance status, disease stage, smoking history (current, for-

mer, never or unknown), ethnicity (white or other),

baseline anemia (yes or no), previous resection (pneumo-

nectomy, lesser resection, or other), previous major

comorbidity (cardiovascular disease, diabetes, both, or

neither), recent surgery (within 4 weeks of study entry),

previous radiotherapy, previous chemotherapy (none,

platinum based, or other), site of disease (lung only, liver

metastases, bone metastases, or other), weight loss (�5%,

<5%, or unknown), Charlson comorbidity index (0 or

1þ), steroids at baseline, anticoagulant use at baseline,

antiplatelet agent use at baseline, hormone use at baseline,

erythropoietin use at baseline, obesity (body mass index

�30 kg/m2 or >30 kg/m2), baseline platelet count (100-

149� 109/L, 150-500� 109/L, or>500� 109/L), and a

previous history of VTE.

Univariate analyses evaluating the association

between each baseline variable and the outcome of interest

were performed using chi-square tests, Fisher exact tests,

or Mantel-Haenszel tests for categoric variables. Continu-

ous variables were compared using analyses of variance.

Multivariate logistic regression was used to deter-

mine which baseline variables were associated independ-

ently with VTE occurrence. A stepwise variable

selection procedure was performed using the minimum

Akaike Information Criterion. For BR.10, because VTE

occurred only in the treatment arm, the multivariate

logistic regression model was fitted only for the treat-

ment arm.

Prognostic Analyses of VTE on Survival

Overall survival (OS) was defined as the time from ran-

domization to the time of death from any cause; other-

wise, it was censored at last date the patient was known to

be alive at the end of the trials. Distributions of OS were

summarized using Kaplan-Meier plots. A Cox propor-

tional hazards model with VTE as a time-dependent cova-

riate was used to study the effect of VTE on OS. The

reported P values along with hazard ratios (HRs) and their

confidence intervals (CIs) were based on the analysis

stratified by treatment arm for the respective trials. Multi-

variate survival analyses were conducted to compare the

VTE and no-VTE groups adjusted for the same baseline

factors that were considered in our multivariate analysis of

factors that were predictive of VTE, excluding a history of

VTE, which was excluded because historic VTE was

included in the definition of ‘‘VTE occurrence’’ (VTE at

any time) that was used for the survival portion of our

analyses.

RESULTS

Baseline Patient Characteristics

Baseline characteristics for the patients in each of the 3 tri-

als are displayed in Table 1. The median age, the propor-

tion of women, and the proportion of patients with

adenocarcinoma were similar for the 3 studies. Differen-

ces in disease stage and performance status were consistent

with differences in study entry criteria. A small number of

patients in each trial had a history of VTE before study

entry. However, in each trial, the number of patients

receiving anticoagulants at the time of study entry

exceeded the number of patients with a history of previous

VTE. Other reasons for anticoagulant use were recorded

only for BR.18 and included atherosclerotic heart disease

(5 patients), peripheral vascular disease (2 patients), atrial

fibrillation (4 patients), stroke (1 patient), and superior

vena caval obstruction (1 patient). The remaining patents

were receiving prophylactic anticoagulants. Few patients

were receiving agents that had a known association with

thrombosis, such as hormones or erythropoietin, at study

entry.

Factors Predictive of VTE

VTE was observed in all trials (Table 2). VTE incidence

ranged from 0% in observation patients on BR.10 to

7.9% overall in BR.18. In BR.10, VTE was associated sig-

nificantly with the administration of adjuvant chemother-

apy (P ¼ .015). Seven patients experienced VTE in the

experimental arm, whereas no patients in the observation

arm developed this complication. On both univariate

analysis and multivariate analysis of the chemotherapy

arm of BR.10, only obesity (P¼ .02) was found to be pre-

dictive of VTE on study.

In BR.18 and BR.21, the occurrence of VTE was in-

dependent of treatment arm; however, the VTE incidence

Original Article

5518 Cancer December 1, 2009

of 7.9%was substantially higher in BR.18 (a trial in which

all patients received chemotherapy) compared with BR.21

(in which the VTE incidence was 2.7%). In multivariate

analysis of BR.18, only a history of VTE (P ¼ .001) was

found to be predictive of VTE on study, whereas no sig-

nificant predictive factors were identified for BR.21.

When BR.18 and BR.21 were combined, a history of pre-

vious VTE (P¼ .0003) remained predictive.

Table 1. Baseline Demographics of Patients in Each National Cancer Institute of Canada Clinical Trial

Percentage of Patients

Trial BR.10, n5482 Trial BR.18, n5774 Trial BR.21, n5731Characteristic Observation,

n5240

CT,

n5242

CT1Placebo,

n5387

CT1MMPI,

n5387

Observation,

n5243

Erlotinib,

n5488

Age, yMedian (range) 61 (34-78) 61 (35-82) 61 (35-83) 61 (35-89) 60 (32-89) 62 (34-87)

£65 67.5 68.2 66.1 65.4 63 61.3

>65 32.5 31.8 33.9 34.6 37 38.7

SexMen 64.2 66.1 73.1 72.9 65.8 64.5

Women 35.8 33.9 26.9 27.1 34.2 35.5

PathologyAdenocarcinoma 53.3 52.9 51.9 49.9 49 50.4

Squamous 37.5 36.8 23.8 22.5 32.1 29.5

Other 9.2 10.3 24.3 27.6 18.9 20.1

ECOG PS0 48.5 49.6 29.5 26.9 14 13.1

1 51.5 50.4 59.2 61 54.3 52.5

2 — — 11.1 12.1 23 25.8

3 — — 0.3 0 8.6 8.6

TNM StageI 45 45.9 — — — —

II 55 54.1 — — — —

IIIA — — 0.3 0 — —

IIIB — — 21.2 21.2 — —

IV — — 78.6 78.8 100 100

BMINot obese 79.9 80.2 86 88 90.9 89.7

Obese 20.1 19.8 14 12 9.1 10.3

Erythropoietin at baselineNo 100 100 100 99.5 98.8 99.2

Yes 0 0 0 0.5 1.2 0.8

Hormone therapy at baselineNo 91.7 88.8 94.1 93 95.5 94.5

Yes 8.3 11.2 5.9 7 4.5 5.5

History of VTENo 98.8 98.8 97.9 95.1 97.9 96.1

Yes 1.3 1.2 2.1 4.9 2.1 3.9

Receiving anticoagulant at baselineNo 96.7 96.7 92 90.2 92.6 91.4

Yes 3.3 3.3 8 9.8 7.4 8.6

Receiving antiplatelet agent at baselineNo 85.4 83.1 88.9 89.4 95.5 93.4

Yes 14.6 16.9 11.1 10.6 4.5 6.6

CT indicates chemotherapy; MMPI, metalloproteinase inhibitor; ECOG PS, Eastern Cooperative Oncology Group performance status; BMI, body mass index;

VTE, venous thromboembolism.

VTE and NSCLC/Hicks et al

Cancer December 1, 2009 5519

Prognostic Effect of VTE at Any Time

on Survival

The OS of the patients who experienced VTE at any time

(past history, at or after randomization), compared with

patients who did not experience VTE, is illustrated in

Figure 1. In BR.18 and BR.21 (the 2 trials in patients with

advanced NSCLC), the Cox regression model with VTE

occurrence as a time-dependent covariate indicated that

VTE was associated with shorter OS (BR.18: HR, 1.61;

95% CI, 1.26-2.07; BR.21: HR, 2.18; 95% CI, 1.57-3.04).

However, VTE at any time was not prognostic in BR.10.

The association between VTE and survival that we

observed on univariate analysis in BR.18 and BR.21 was

explored further after adjusting for other important base-

line factors. In both BR.18 and BR.21, VTE occurrence

was associated independently with shorter OS. In BR.18,

the HR for death if VTE occurred was 1.35 (95% CI,

1.04-1.75; P ¼ .026). In BR.21, the HR for death if

VTE occurred was 2.128 (95% CI, 1.50-2.98; P <

.0001). When the trials were pooled, the HR was 1.60

(95% CI, 1.30-1.98; P < .0001) after adjusting for other

known prognostic factors. Causes of death for patients

Table 2. Association Between Venous Thromboembolic Event Occurring While on Study andBaseline Factors

P

Baseline Factor No.Evaluable

No.With VTE

UnivariateAnalysis

MultivariateAnalysis

Trial BR.10 7 (1.5%)

Treatment

Chemotherapy 242 7 .015 NA

Observation 240 0

Obesity*

Obese 48 4 .03 .02

Not obese 194 3

Trial BR.18 61 (7.9%)

Treatment

CT 387 28 .51 NS

CT1MMPI 387 33

History of VTE

Yes 27 7 .0001 .001

No 747 54

Sex

Men 565 39 .097 NS

Women 209 22

Anticoagulant

Yes 69 12 .002 NS

No 705 49

Erythropoietin

Yes 2 1 .04 .08

No 772 60

Trial BR.21 20 (2.7%)

Treatment

Erlotinib 488 13 .87 NS

Placebo 243 7

History of VTE

Yes 24 2 .08 NS

No 707 18

Ethnicity

White 567 19 .058 NS

Other 164 1

VTE indicates venous thromboembolism; NA, not applicable because model fitted for chemotherapy arm only; CT, chem-

otherapy; MMPI, metalloproteinase inhibitor; NS, not significant.* The analysis of obesity was completed by using the experimental arm only.

Original Article

5520 Cancer December 1, 2009

who had VTE either at study entry or after randomiza-tion are listed in Table 3.

DISCUSSION

Our analyses suggested that the risk of VTE is increased in

patients with NSCLC at all stages who receive chemother-

apy whether it is given as postoperative adjuvant therapy

or for the treatment of advanced disease. We also observed

that patients with advanced NSCLC also have a higher

risk of VTE than patients with early stage disease, even if

they are not receiving cytotoxic therapy, as demonstrated

by the 2.7% incidence in BR.21 placebo-treated patients

compared with the 0% incidence in the observation arm

of BR.10. Although it is possible that investigators did not

report VTE events for patients who were randomized to

the observation arm of BR.10 because of less frequent

follow-up compared with patients on chemotherapy who

were seen weekly, it appears that adjuvant chemotherapy

for early stage NSCLC is associated with an increased risk

of nonfatal VTE. This result is consistent with reports

from breast cancer trials in which both adjuvant chemo-

therapy and hormone therapy were correlated with an

increased risk of VTE.5-7

The results of the current study also are in keeping

with the increased risk of VTE in patients with advanced

lung cancer who received chemotherapy reported by Khor-

ana et al.12 How chemotherapy contributes to VTE risk has

not to our knowledge been elucidated completely but likely

involves a combination of direct endothelial damage and

down-regulation of endogenous anticoagulants.17 Other

drugs that are used frequently in both cancer and noncancer

patients, particularly hormones18,19 and erythropoietin,20-22

are associated with an increased risk of VTE in the malig-

nant setting.5-7 In the current series, few patients were

receiving hormone therapy, most likely because of the na-

ture of the lung cancer population; therefore, we did not

identify an association with the use of oral contraceptives or

hormone-replacement therapy. Similarly, only 2 patients

received with erythropoietin; however, it is interesting to

note that 1 of those patients did experience VTE.

Although the patients randomized to BR.18, all of

whom received chemotherapy and had advanced NSCLC,

had the highest incidence of VTE, the addition of the

metalloproteinase inhibitor BMS-275291 did not

increase the risk. This result is in contrast to that observed

FIGURE 1. Overall survival is illustrated by venous throm-

boembolism (VTE) status in (A) National Cancer Institute of

Canada (NCIC) Clinical Trial BR.10, a randomized study of the

postoperative combination of vinorelbine and cisplatin versus

observation in patients with stage IB/II nonsmall cell lung

cancer (NSCLC); (B) NCIC Clinical Trial BR.18, a randomized

study of the combination of paclitaxel and carboplatin with

or without the metalloproteinase inhibitor BMS-275291 in

patients with advanced NSCLC; and (C) NCIC Clinical Trial

BR.21, a randomized study of erlotinib versus placebo in pre-

viously treated patients with NSCLC.

VTE and NSCLC/Hicks et al

Cancer December 1, 2009 5521

by Behrendt and Ruiz,13 who reported an increased risk of

VTE when 15 mg of the metalloproteinase inhibitor pri-

nomastat was added to platinum-based chemotherapy.

The lack of any association of VTE with BMS-275291 in

our analysis of BR.18 suggests that the increased incidence

of VTE observed with prinomastat may not be a class

effect. Alternatively, the absence of an association between

VTE and metalloproteinase use in the current study may

reflect a dose effect. Behrendt and Ruiz13 did not report

an increased incidence of VTE when lower doses of prino-

mastat were administered. In BR.18, only a single dose

level (1200 mg) of BMS-275291 was tested.

The 2.7% rate of VTE in patients with very

advanced NSCLC in BR.21 was intermediate between

that of the patients in BR.10 who did not receive adjuvant

chemotherapy (0%) and that observed in BR.18 (7.9%),

in which all patients received chemotherapy. The addition

of erlotinib did not appear to increase the risk of VTE. To

our knowledge to date, there have been few published

reports of an increased risk of VTE with erlotinib. Simi-

larly, sunitinib and sorafenib (also tyrosine kinase inhibi-

tors that target the vascular endothelial growth factor

[VEGF] receptor) have not been associated with high rates

of VTE in early trials.23 However, semaxibin (another ty-

rosine kinase inhibitor that targets VEGF) has been asso-

ciated with moderate-to-high VTE rates, particularly

when combined with chemotherapy.23,24 In addition, a

growing body of literature suggests that bevacizumab, a

recombinant monoclonal antibody against VEGF, may

increase the risk of VTE.25 It is possible that the thrombo-

genecity of antiangiogenic agents is not a class effect and

depends on the specific mechanism of action of each drug.

Alternatively, the propensity of antiangiogenic agents to

increase VTE incidence may depend on tumor type,

tumor stage, and whether the antiangiogenic agent is

given as monotherapy or in combination with other

agents. In the current study, erlotinib was administered as

monotherapy to patients with previously treated NSCLC.

We observed that obesity was predictive for VTE in

patients with early stage NSCLC. The association

between obesity and VTE is well recognized in the general

population,17-19 and a similar association also has been

reported in cancer patients.19 Although the cause is not

known, it appears reasonable to postulate that both

decreased mobility and increased endogenous estrogen

production may contribute to an increased risk of VTE in

obese patients. Although obesity was predictive for VTE

in BR.10, we were unable to confirm similar effects in

BR.18 and BR.21.

In BR.18, predictive factors for VTE included a pre-

vious history of VTE, which is a well recognized risk fac-

tor for subsequent VTE in other settings, particularly

postoperatively.21 This is important, because it suggests

that it may be possible to identify patients with advanced

NSCLC at particularly high risk of VTE who may benefit

from VTE prophylaxis. In support of this is our observa-

tion that patients who were receiving anticoagulation did

not have a higher risk of VTE, even when they were

receiving this treatment for previous thrombotic events.

An interesting observation was that on-study VTE

was more likely to be fatal in BR.18 and BR.21 compared

with BR.10. The reasons for this are unclear. Episodes of

sudden death or sudden dyspnea and death in patients

with advanced NSCLC sometimes are attributed to VTE,

although postmortem confirmation of this usually is not

undertaken. It is possible, therefore, that some of the fatal

VTEs that were reported in BR.18 and BR.21 were sud-

den cardiac or neurologic events unrelated to VTE.

VTE at any time was prognostic in BR.18 and

BR.21. Although this may be explained in part by the

Table 3. Causes of Death for Patients With Venous Thromboembolism at Any Time

No. of PatientsWith VTE*

Non-VTE–RelatedDeaths

Deaths From VTE

Trial BR.10: 10 patients with VTEy 3 0

Trial BR.18: 81 patients with VTEy 62 Seven deaths from pulmonary embolus and 1

death from superior vena cava thrombosis

Trial BR.21: 42 patients with VTEy 30 Eight deaths from pulmonary embolus

VTE indicates venous thromboembolism.* Included VTE that occurred at any time.yNot all patients with VTE had died at the time of study analysis.

Original Article

5522 Cancer December 1, 2009

incidence of fatal VTE, an increased risk of non-VTE

death cannot be ruled out. An association between VTE

occurrence and increased cancer mortality has been

reported in cancer patients receiving chemotherapy:

Kuderer et al26 demonstrated that the HR for death

among 4458 patients with various malignancies was 4.90

(95% CI, 2.27-10.60; P < .001) in patients who had

experienced VTE. Pulmonary embolism was the cause of

death for 5.4% of patients, which suggests that, similar to

the current study, the shorter survival cannot be explained

by VTE events alone. It has been postulated that this

increase in the death rate may be caused by the up-regula-

tion of endogenous procoagulant factors in malignancy.27

In addition to being prothrombotic, there is preclinical

evidence that procoagulant factors, such as tissue factor

and thrombin, may contribute to angiogenesis, which, in

theory, could potentiate cancer growth.28,29 An elevation

of other nonspecific markers of coagulation and fibrinoly-

sis also has been noted frequently in cancer.30 For exam-

ple, 1 study of patients with lung cancer noted a

correlation between elevated d-dimer levels and poor sur-

vival, even after controlling for other recognized prognos-

tic predictors.31 Recently, genetic polymorphisms in

selectins have been associated with VTE in cancer

patients.32

There have been several trials of thromboprophylaxis

in cancer patients. Particularly with respect to lung cancer,

the TOPIC II trial, which compared the low-molecular-

weight heparin (LMWH) certoparin with placebo in 547

patients with advanced NSCLC, demonstrated a nonsigni-

ficant reduction in the risk of VTE (4.5% vs 8.3%; P ¼.07).33 That study has not reported survival results, and the

full publication is awaited. Both the American Society for

Clinical Oncology34 and the National Comprehensive

Cancer Network35 recommend the use of prophylactic anti-

coagulation with LMWH for hospitalized patients with

cancer in view of the high rate of VTE in this population.36

Neither guideline goes so far as recommending prophylactic

anticoagulation for ambulatory patients who are receiving

chemotherapy, although there is now mounting evidence to

suggest that such patients are at increased risk for VTE.

Khorana et al37 have developed a predictive model for VTE

in patients who are receiving chemotherapy. Patients in

their high-risk group had VTE rates of approximately 7%

in both their derivation and validation cohorts. This rate is

nearly identical to the VTE rates we report here in the

BR.18 trial of chemotherapy in patients with advanced

NSCLC.

The current study had several limitations. The analy-

ses were conducted retrospectively. Because VTE incidence

was not a prospectively planned endpoint of these trials, it

is possible that some events of VTE were not reported. In

addition, objective diagnostic criteria for VTE were not pre-

specified or required of participating centers. However, we

did attempt to verify all VTE events by manually reviewing

all enrollment data and all adverse event reports. We

believe, however, that most of the events that we classified

as such were indeed VTE in view of the finding that the

rate of VTE in BR.18 (7.9%) was nearly identical to the

8.3% rate reported in the TOPIC II trial, in which there

was greater verification of the diagnosis because VTE was

the primary endpoint of the study.33

In summary, we believe the current study is the largest

review of VTE reported in NSCLC to date. We have dem-

onstrated that VTE is a relatively common occurrence in

individuals with lung cancer, particularly in patients with

advanced NSCLC and those who receive chemotherapy.

We have also demonstrated that obesity and a history of

VTE may increase the risk of VTE. It is noteworthy that

VTE at any time was identified as a prognostic factor for

patients with advanced NSCLC. The possibility that

manipulation of the hemostatic balance may influence not

only VTE occurrence but also survival is enticing and is

being explored in several lung cancer trials. These ongoing,

randomized controlled trials assessing treatment with

LMWH versus no anticoagulation in patients with various

stages of lung cancer may provide a definitive answer with

regard to the efficacy of VTE prophylaxis and may shed fur-

ther light on whether there is a survival benefit with such

treatment. In the interim, patients who are receiving chemo-

therapy for NSCLC should be counseled on the increased

risk of VTE in this setting and on how to recognize the

symptoms of VTE.

Conflict of Interest Disclosures

Supported in part by grants from the Canadian Cancer Society tothe National Cancer Institute of Canada Clinical Trials Group.

References

1. Heit JA, Silverstein MD, Mohr DN, et al. Risk factors fordeep vein thrombosis and pulmonary embolism: a popula-tion-based case-control study. Arch Intern Med. 2000;160:809-815.

VTE and NSCLC/Hicks et al

Cancer December 1, 2009 5523

2. Levitan N, Dowlati A, Remick SC, et al. Rates of initialand recurrent thromboembolic disease among patients withmalignancy versus those without malignancy. Risk analysisusing Medicare claims data. Medicine (Baltimore). 1999;78:285-291.

3. Rickles FR, Levine M, Edwards RL. Haemostatic altera-tions in cancer patients. Cancer Metastasis Rev. 1992;11:237-248.

4. Maiolo A, Tua A, Grignani G. Hemostasis and cancer:tumour cells induce the expression of tissue factor-like pro-coagulant activity on endothelial cells. Haematologica. 2002;87: 624-628.

5. Saphner T, Tormey DC, Gray R. Venous and arterialthrombosis in patients who received adjuvant therapy forbreast cancer. J Clin Oncol. 1991;9:286-294.

6. Clahsen PC, van de Velde CJ, Julien JP, et al. Throm-boembolic complications after perioperative chemotherapyin women with early breast cancer: a European Organiza-tion for Research and Treatment of Cancer Breast CancerCooperative Group study. J Clin Oncol. 1994;12:1266-1271.

7. Levine MN, Gent M, Hirsh J, et al. The thrombogeniceffect of anticancer drug therapy in women with stage IIbreast cancer. N Engl J Med. 1988;318:404-407.

8. Zangari M, Anaissie E, Barlogie B, et al. Increased risk ofdeep-vein thrombosis in patients with multiple myelomareceiving thalidomide and chemotherapy. Blood. 2001;98:1614-1615.

9. Palumbo A, Bringhen S, Caravita T, et al. Oral melphalanand prednisone chemotherapy plus thalidomide comparedwith melphalan and prednisone alone in elderly patientswith multiple myeloma: randomised controlled trial. Lancet.2006;367:825-831.

10. Rus C, Bazzan M, Palumbo A, et al. Thalidomide in frontline treatment in multiple myeloma: serious risk of venousthromboembolism and evidence for thromboprophylaxis.J Thromb Haemost. 2004;2:2063-2065.

11. Chew HK, Davies AM, Wun T, et al. The incidence ofvenous thromboembolism among patients with primarylung cancer. J Thromb Haemost. 2008;6:601-608.

12. Khorana AA, Francis CW, Culakova E, et al. Risk factorsfor venous thromboembolism in a prospective observationalstudy. Cancer. 2005;104:2822-2829.

13. Behrendt CE, Ruiz RB. Venous thromboembolism amongpatients with advanced lung cancer randomized to prinoma-stat or placebo, plus chemotherapy. Thromb Haemost.2003;90:734-737.

14. Winton T, Livingston R, Johnson D, et al. Vinorel-bine plus cisplatin versus observation in resected non-small-cell lung cancer. N Engl J Med. 2005;352:2589-2597.

15. Leighl NB, Paz-Ares L, Douillard JY, et al. Randomizedphase III study of matrix metalloproteinase inhibitor BMS-275291 in combination with paclitaxel and carboplatin in

advanced non-small-cell lung cancer: National Cancer Insti-tute of Canada-Clinical Trials Group Study BR.18. J ClinOncol. 2005;23:2831-2839.

16. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, et al. Erloti-nib in previously treated non-small-cell lung cancer. N EnglJ Med. 2005;353:123-132.

17. Heit JA. Risk factors for venous thromboembolism. ClinChest Med. 2003;24:1-12.

18. Abdollahi M, Cushman M, Rosendaal FR. Obesity: risk ofvenous thrombosis and the interaction with coagulationfactor levels and oral contraceptive use. Thromb Haemost.2003;89:493-498.

19. Pomp ER, le Cessie S, Rosendaal FR, et al. Risk of venousthrombosis: obesity and its joint effect with oral contracep-tive use and prothrombotic mutations. Br J Haematol.2007;139:289-296.

20. Drueke TB, Locatelli F, Clyne N, et al. Normalizationof hemoglobin level in patients with chronic kidneydisease and anemia. N Engl J Med. 2006;355:2071-2084.

21. Singh AK, Szczech L, Tang KL, et al. Correction of anemiawith epoetin alfa in chronic kidney disease. N Engl J Med.2006;355:2085-2098.

22. Wright JR, Ung YC, Julian JA, et al. Randomized, double-blind, placebo-controlled trial of erythropoietin in non-small-cell lung cancer with disease-related anemia. J ClinOncol. 2007;25:1027-1032.

23. Elice F, Jacoub J, Rickles FR, et al. Hemostatic complica-tions of angiogenesis inhibitors in cancer patients. Am JHematol. 2008;83:862-870.

24. Kuenen BC, Rosen L, Smit EF, et al. Dose-finding andpharmacokinetic study of cisplatin, gemcitabine, andSU5416 in patients with solid tumors. J Clin Oncol.2002;20:1657-1667.

25. Nalluri SR, Chu D, Keresztes R, et al. Risk of venousthromboembolism with the angiogenesis inhibitor bevacizu-mab in cancer patients: a meta-analysis. JAMA.2008;300:2277-2285.

26. Kuderer NM, Francis CW, Culakova E, et al. Venousthromboembolism and all-cause mortality in cancer patientsreceiving chemotherapy [abstract]. J Clin Oncol.2008;26(15suppl):507s. Abstract 9521.

27. Hicks LK, Selby R. Recent advances in the epidemiology,pathophysiology, and management of thrombosis in cancerpatients. Lab Med. 2004;35:492-496.

28. Rickles FR, Patierno S, Fernandez PM. Tissue factor,thrombin, and cancer. Chest.2003;124:58S-68S.

29. Versteeg HH, Peppelenbosch MP, Spek CA. Tissue factorsignal transduction in angiogenesis. Carcinogenesis.2003;24:1009-1013.

30. Beer JH, Haeberli A, Vogt A, et al. Coagulation markerspredict survival in cancer patients. Thromb Haemost.2002;88:745-749.

Original Article

5524 Cancer December 1, 2009

31. Buccheri G, Torchio P, Ferrigno D. Plasma levels ofD-dimer in lung carcinoma: clinical and prognostic signifi-cance. Cancer. 2003;97:3044-3052.

32. Grau JJ, Font C, Visa L, et al. Cancer and venous throm-boembolism: role of genetic polymorphisms in selectins andtissue factor [abstract]. J Clin Oncol.2008;26(15 suppl):503s. Abstract 9504.

33. Haas SK, Kakkar AK, Kemkes-Matthes B, et al. Preventionof venous thromboembolism with low-molecular-weightheparin in patients with metastatic breast or lung cancer—results of the TOPIC studies [abstract]. J Thromb Haemost.2005;3(suppl 1). Abstract OR059.

34. Lyman GH, Khorana AA, Falanga A, et al. AmericanSociety of Clinical Oncology guideline: recommendations

for venous thromboembolism prophylaxis and treatmentin patients with cancer. J Clin Oncol. 2007;25:5490-5505.

35. Khorana AA. The NCCN clinical practice guidelines onvenous thromboembolic disease: strategies for improvingVTE prophylaxis in hospitalized cancer patients. Oncologist.2007;12:1361-1370.

36. Khorana AA, Francis CW, Culakova E, et al. Throm-boembolism in hospitalized neutropenic cancer patients.J Clin Oncol. 2006;24:484-490.

37. Khorana AA, Kuderer NM, Kulakova E, et al.Development and validation of a predictive model forchemotherapy-associated thrombosis. Blood. 2008;111:4902-4907.

Cancer December 1, 2009 5525

VTE and NSCLC/Hicks et al