Thrombosis Research 129, Supplement 1 (2012) S10–S15

PL-02

Risk assessment for cancer-associated thrombosis: What is the best approach?

Alok A. Khorana *James P. Wilmot Cancer Center, and the Department of Medicine, University of Rochester, Rochester, NY, USA

A R T I C L E I N F O A B S T R A C T

Keywords:Venous thromboembolismPreventionRisk factorsCancer

Venous thromboembolism (VTE) is an increasingly frequent complication of cancer and its treatments. Onein five cancer patients are estimated to develop venous and arterial events during the natural history of theirillness. However, the risk for VTE varies widely between various subgroups of cancer patients and even inthe same cancer patient over time. This narrative review focuses on risk factors, biomarkers and risk as-sessment tools and attempts to clarify approaches to risk stratification. Clinical risk factors include primarysite of cancer, chemotherapy, anti-angiogenic therapy, surgery and hospitalization. Predictive and candidatebiomarkers include platelet and leukocyte counts, hemoglobin, D-dimer and tissue factor. However, singlerisk factors or biomarkers have not, in general, been able to identify sufficiently high-risk populations. A clin-ical risk score, incorporating 5 simple clinical and laboratory variables, has now been studied in over 10,000patients and can successfully categorize patients at low- and high-risk for VTE. Recent trials have shown thatoutpatient prophylactic anticoagulation is both safe and effective, but event rates have been highly variable.Targeted thromboprophylaxis provides an optimal risk–benefit ratio and the best opportunity to reduce theburden of VTE and its consequences for patients with cancer.

© 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Cancer patients frequently experience thromboembolic events(TEEs) and the risk of such events is increased several-fold inthe cancer population when compared to the general population[1]. Additionally, the frequency of both venous and arterial eventshas increased substantially in the past decade [2]. However, therisk varies widely between subgroups of cancer patients and inindividual patients over time [3,4]. Even if we are to accept theconventional estimate that one in five cancer patients will developvenous thromboembolism (VTE) during the natural history of theirillness, this implies that four in five cancer patients will not sufferfrom VTE. VTE is largely preventable with the appropriate useof prophylactic anticoagulants [5]. However, unlike other settingswhere prophylactic anticoagulation is successful and widely used,the risk for VTE in cancer can persist over a period of severalweeks to months. Identifying cancer patients at highest risk for VTE,therefore, is of crucial importance to avoid the burden and potentialcomplications of long-term anticoagulants in patients who do notneed them and, conversely, to provide these drugs to patients trulyat risk. This narrative review will focus on reviewing recent dataon clinical risk factors, predictive biomarkers and risk assessmenttools with an emphasis on identifying best practice approaches toassessing risk of VTE in the cancer patient.

* Correspondence to: Alok A. Khorana, MD, 601 Elmwood Ave, Box 704,Rochester, NY 14642, USA. Tel.: +1 (585) 273-4150; fax: +1 (585) 273-1042.

E-mail address: alok_khorana@urmc.rochester.edu (A.A. Khorana).

0049-3848/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.

2. A note on “rates” of TEE

A host of publications in the past decade have provided con-temporary information regarding the incidence, prevalence andburden of TEEs in the cancer population (reviewed in [3]). However,rates of VTE reported across these studies vary widely even whensimilar populations are studied. This has been a source of frus-tration for clinicians and investigators focused on study design ofprophylaxis studies, since sample size determinations rest heavilyon anticipated VTE rates. It is important, therefore, to understandthe context within which rates are reported and discussed in theliterature as well as the reasons for these disparities in reportedrates (Table 1). Firstly, the study endpoint varies across studies;more recent studies utilize what has been described [6] as an“extended” definition of TEE, which includes visceral thromboses,arterial events and incidentally discovered events. There is goodscientific rationale to support such an extended definition sincethrombotic events are linked pathophysiologically and because in-cidental and similar adverse outcomes are noted for visceral eventsas for symptomatic events [7,8]. However, such a definition needsto be standardized. Secondly, rates are generally higher if TEEconstitute the primary or secondary endpoint of the study and TEErates are often underestimated when events are recorded as toxicitydata for clinical trials. Indeed, in one analysis of a clinical trial inadvanced colorectal cancer, 90% of VTE cases were not identifiedby usual toxicity reporting and only discovered on subsequentchart review [9]. In a meta-analysis of TEE endpoints in cancerstudies, incidence rates of VTE outcomes were 3–55 times higherfor active surveillance than for passive surveillance [10]. Next, the

A.A. Khorana / Thrombosis Research 129, Supplement 1 (2012) S10–S15 S11

Table 1Reasons for disparities in rates of TEEs across studies.

Study endpointsHigher rates if incidental events includedHigher if visceral thromboses includedHigher if arterial events includedHigher if screening studies conductedHigher if TEE primary or secondary endpoint; lower if TEE collected as partof toxicity reporting

Higher for active versus passive surveillance

Study populationHigher rates in North America; lower in AsiaHigher in retrospective studies; lower in clinical trialsHigher if studies focused on high-risk populations (on systemic therapy,specific sites of cancer); lower in large, heterogeneous study populations

Higher in more contemporaneous reports

Duration of follow-up

composition of the study population can greatly affect reportedrates since the “case mix” is in and of itself an important riskfactor for VTE. Thus, studies that are population-based and involveheterogeneous cancer sites generally report lower rates than stud-ies focused on specific cancers (e.g., pancreas or lung [11,12]) orspecific chemotherapy regimens (e.g., cisplatin-based therapy [13]).Finally, the duration of follow-up can influence rates as well. Anexcellent illustration of such disparities is the difference in ratesreported for high-risk patients in the development and validationof a clinical risk score. In the original report [14], high-risk patientshad a risk of VTE of 6.7–7.1% but this was based on a medianfollow-up of approximately 2.5 months; in addition, VTE reportingwas not a primary or secondary endpoint of the study. In anexternal validation of this risk score performed by the Vienna CATSgroup [15], rates in the high-risk cohort were much greater at 17.7%but this was reported at a 6 month follow-up period and VTE wasin fact a primary endpoint of this study. It is important to keepall of these considerations in mind when conducting cross-studycomparisons and when utilizing these data to estimate rates for thedesign of future studies.

3. Clinical risk factors

Clinical risk factors for VTE include patient-related, cancer-related and treatment-related risk factors (Table 2). Patient-relatedrisk factors include demographics such as older age and African–American race [2]. Comorbid conditions are strongly associatedwith VTE, particularly infection, pulmonary disease, renal diseaseand obesity [16]. Cancer patients with a prior history of VTE havea 6–7 fold increased risk of developing VTE compared to cancerpatients with no history of VTE [17]. The primary site of cancerhas historically been considered the most important clinical riskfactor. Highest rates of VTE are observed in patients with brain,pancreas, stomach, kidney, ovary and lung cancers [3]. Patients withhematologic malignancies are also at high risk (in one study, oddsratio [OR] 28, 95% CI 4.0–199.7) [4]. The rate of VTE is especiallyhigh during the initial period after diagnosis. In the population-based study discussed above, risk of VTE was highest in the first 3months after initial diagnosis of cancer (OR 53.5, 95% CI 8.6–334.3),with some degree of elevated risk persisting for several years [4].

Therapeutic interventions further enhance the risk of VTE. Theuse of systemic chemotherapy is associated with a 2- to 6-foldincreased risk of VTE compared to the general population [18,19].VTE risk may further be influenced by specific anti-neoplasticagents and regimens. For instance; in an analysis of patientswith metastatic gastric and gastro-esophageal junction cancers,thromboembolism rates were 15.1% in patients receiving cisplatincompared to 7.6% in patients receiving oxaliplatin [20]. Anti-angiogenic agents, particularly thalidomide and lenalidomide, have

Table 2Selected clinical risk factors and biomarkers for cancer-associated thrombosis.

Patient-associated risk factorsOlder ageRaceGenderMedical comorbiditiesObesityPast history of thrombosis

Cancer-associated risk factorsPrimary siteStageCancer histology (higher for adenocarcinoma than squamous cell)Time after initial diagnosis (highest in first 3–6 months)

Treatment-associated risk factorsChemotherapyAnti-angiogenic agents (thalidomide, lenalidomide)Hormonal therapyErythropoiesis-stimulating agentsTransfusionsIndwelling venous access devicesRadiationSurgery

BiomarkersCurrently widely availablePlatelet count (≥350,000/mm3)Leukocyte count (>11,000/mm3)Hemoglobin (<10 g/dL)D-dimer

Investigational or not widely availableTissue factor (antigen expression, circulating microparticles, antigen oractivity levels)

Soluble P-selectin (>53.1 ng/mL)Factor VIIIProthrombin fragment F 1 +2 (>358 pmol/L)

been associated with high rates of VTE when given in combinationwith dexamethasone or chemotherapy. Bevacizumab-containingregimens have been associated with increased risk for an arterialthromboembolic event (hazard ratio [HR] 2.0, 95% CI 1.05–3.75) butthe data for risk of VTE are conflicting [21,22]. The anti-angiogenicmultiple tyrosine kinase inhibitors sunitinib and sorafenib, havealso been associated with elevated risk for arterial events [RR 3.03(95% CI, 1.25 to 7.37)] but not for VTE [23]. The use of centralvenous access devices adds to this risk, although recent studieshave demonstrated an overall low risk with contemporary catheters[24]. Supportive care agents can also add to the risk For instance,erythropoietin and darbopoetin are associated with a significantlyincreased risk of VTE (RR = 1.57; 95% CI, 1.31 to 1.87) and mortality(HR = 1.10; 95% CI, 1.01–1.20) [25]. Cancer patients undergoingsurgery have a two-fold increased risk of postoperative VTE ascompared to non-cancer patients; elevated risk can persist for upto 7 weeks [26]. Hospitalization substantially increases the risk ofdeveloping VTE in cancer patients (OR 2.34, 95% CI 1.63–3.36) [27].

4. Biomarkers

Recent studies have identified several predictive and candidatebiomarkers (Table 2). Baseline (i.e., pre-chemotherapy) elevatedplatelet and leukocyte counts, and low hemoglobin levels have allbeen demonstrated to be risk factors for chemotherapy-associatedVTE [14,20]. In a prospective study of cancer patients receivingchemotherapy, rates of VTE were 4% in patients with elevatedplatelet counts as compared to 1.3% for those without (adjustedOR 2.8, 95% CI 1.6–5) [28]. The Vienna Cancer and ThrombosisStudy (CATS) also found high platelet count to be an independentrisk factor for VTE (HR = 3.50; 95% CI, 1.52 to 8.06, p = 0.0032)in 665 cancer patients undergoing chemotherapy. Patients withplatelet counts in the 95th percentile or above (platelet count of443×109/L) had a 34.3% cumulative probability of developing VTE

S12 A.A. Khorana / Thrombosis Research 129, Supplement 1 (2012) S10–S15

in contrast to only 5.9% in the group with counts below the des-ignated threshold [29]. Similarly in a prospective analysis of 4,405ambulatory cancer patients initiating chemotherapy VTE occurredin 25 of 561 patients (4.5%) with baseline leukocytosis compared to68 of 3,830 (1.8%) without leukocytosis (p <0.0001) [30].

D-dimer is also a widely studied biomarker that is also widelyavailable clinically. In colorectal cancer, patients with elevatedD-dimer (defined as >0.3 mg/L) had a 20% (95% CI, 12 to 31%)one-year incidence of DVT versus 5% (95% CI, 2 to 12%) for otherpatients (adjusted HR 6.53; 95% CI, 1.58 to 27.0) [31]. ElevatedD-dimer in addition to prothrombin split products (as defined by acutoff set at the 75th percentile of the total study population) wereassociated with increased risk of VTE (HR = 1.8; 95% CI, 1.0 to 3.2; p= 0.048 and HR = 2.0; 95% CI, 1.2 to 3.6; p = 0.015 respectively) inthe Vienna CATS registry [32].

Tissue factor (TF), the physiologic initiator of hemostasis, iswidely expressed across a variety of human malignancies [33].Unfortunately, there is currently no consensus on a “standard” assayto evaluate TF. Reports initially suggested a significant associationof elevated TF with subsequent VTE; for instance, a retrospectiveanalysis revealed a 1-year cumulative incidence of VTE of 34.8%in patients with tissue factor-bearing microparticles versus 0% inthose without detectable tissue factor-bearing microparticles (p =0.002) [34]. However, the majority of data have been derived frompatients with certain cancers, particularly pancreas [35] and ovary[36]. Initial reports suggested a significant association of elevated TFwith subsequent VTE [34,37]. More recently, in a recent large studyof cancer patients with a heterogeneous mix of cancer patients,elevated procoagulant microparticles (not TF-specific) were notfound to be predictive of VTE [38]. Further, in a prospective analysisof subgroups of the Vienna CATS registry, TF was predictive ofVTE in pancreatic but not brain or colorectal cancers [39]. TF musttherefore still be considered an investigational biomarker awaitingstandardization prior to clinical use, with potential value in selectmalignancies, such as pancreatic cancer.

5. Risk assessment tools

Oncology treatment paradigms are moving toward individual-ization or “personalization” of therapy. Such a therapeutic approachacknowledges that risk for various cancer-related outcomes andcomplications varies significantly between individual patients andthat a “one size fits all” approach does not benefit individual pa-tients. In supportive oncology, for instance, prophylactive myeloidgrowth factors are only recommended if the risk of febrile neu-tropenia is estimated to be 20% or higher. Such an approach hastwo major advantages: it allows targeting of patients truly at risk,and it reduces adverse effects and resource utilization in low-riskpatients.

As is evident from the preceding discussion of risk factors andbiomarkers, cancer-associated VTE is a multi-factorial illness. Aclinical risk score can identify cancer patients at high-risk for VTEby utilizing a combination of easily available clinical and laboratoryvariables (Table 3) [14]. The risk score for VTE was derived from

Table 3Predictive model for chemotherapy-associated VTE [14].

Patient characteristics Risk score *

Site of cancerVery high risk (stomach, pancreas) 2High risk (lung, lymphoma, gynecologic, bladder, testicular) 1

Prechemotherapy platelet count 350,000/mm3 or more 1Hemoglobin level less than 10 g/dl or use of red cell growth factors 1Prechemotherapy leukocyte count more than 11,000/mm3 1Body mass index 35 kg/m2 or more 1

* Risk scores: High risk ≥3; intermediate risk = 1–2; low risk = 0.

a development cohort of 2,701 patients and then validated in anindependent cohort of 1,365 patients from a prospective registry.The stage-adjusted multivariate model identified five predictivevariables. Observed rates of VTE in the development and validationcohorts were 0.8% and 0.3% in the low-risk category, 1.8% and 2%in the intermediate-risk category and 7.1 and 6.7% in the high-riskcategory, respectively. This model was initially externally validatedby the prospective Vienna CATS study in 819 cancer patients [40].The 6-month cumulative probabilities of developing VTE in thisstudy population were 1.5% (score of 0), 3.8% (score of 1), 9.4%(score of 2) and 17.7% (score ≥3). Multiple other retrospective andprospective studies have further validated this Risk Score, includingin populations quite different from the original cohort (Table 4).Mandalà and colleagues evaluated risk factors for VTE in 1,415patients enrolled in phase I studies conducted in Southern Europe[41]. Fifty-six patients (4%) developed a VTE. At univariate analysis,the risk score, the combination of an antiangiogenic agent witha cytotoxic drug, and the time from first cancer diagnosis wereassociated with VTE. The multivariate analysis, however, confirmedonly a statistically significant association for the risk score. Thehazard ratio of VTE occurrence was 7.88 (95% CI 2.86–21.70) and2.74 (95% CI 1.27–5.92) for the high (≥3) and intermediate (1–2)scores as compared with low-risk (score = 0). Similarly, in 932patients treated with cisplatin-based chemotherapy for any type ofmalignancy at Memorial Sloan-Kettering Cancer Center in 2008, therisk score remained significantly associated with TEE after adjustingfor multiple variables, including performance status [13]. Includingall studies that have evaluated efficacy of this risk score, it has nowbeen studied or validated in over 10,000 patients.

Furthermore, the Vienna group has described expansion of thisoriginal risk score with the inclusion of two additional biomarkers:D-dimer and soluble P-selectin. In the expanded risk model, thecumulative VTE probability after 6 months in patients with thehighest score (≥5, n = 30) was 35.0% and 10.3% in those with anintermediate score (score 3, n = 130) as opposed to only 1.0% inpatients with score 0 (n = 200). This expanded risk score, whilepromising, requires further validation in other studies. Use of theexpanded score is further limited by the lack of wide availability ofthe P-selectin assay and the small number of patients at the highestrisk levels.

Patients with myeloma are at very high risk for VTE, particularlywhen treated with specific regimens. A risk assessment algorithmhas recently been proposed by the International Myeloma WorkingGroup [42]. Aspirin, warfarin or LMWH are recommended based onthe level of risk identified by this algorithm. It is important to notethat this risk assessment tool is based on expert consensus and hasnot been validated.

6. Lessons from prophylaxis studies

Multiple randomized controlled trials (RCTs) and meta-analyseshave demonstrated the safety and efficacy of anticoagulants inreducing the incidence of VTE in several high-risk settings. Inthe cancer population, recent studies have focused on canceroutpatients receiving chemotherapy although thromboprophylaxisis largely only recommended in the inpatient setting.

6.1. Hospitalized medical cancer patients

Three large RCTs in acutely ill medical patients have demon-strated reduced rates of VTE with the use of prophylactic LMWHor fondaparinux [43–45]. Unfortunately, no cancer-specific RCTshave been conducted and in the medical studies, cancer patients(including those with previous history of cancer) represented onlya small minority (5–15%) of the study population. Despite thislack of cancer-specific evidence, current guidelines recommend

A.A. Khorana / Thrombosis Research 129, Supplement 1 (2012) S10–S15 S13

Table 4Rates of VTE according to risk score in various studies.

Study Type, f/u N Low risk Intermediate risk High risk(score = 0) (score = 1–2) (score ≥3)

Khorana et al. [14], 2008 Development cohort, 2.5 mos 2,701 0.8% 1.8% 7.1%Khorana et al. [14], 2008 Validation cohort, 2.5 mos 1,365 0.3% 2% 6.7%Kearney et al. [52], 2009 Retrospective, 2 yrs 112 5% 15.9% 41.4%Price et al. [53], 2010 Retrospective, pancreatic, NA 108 – * 14% 27%Ay et al. [40], 2010 Prospective, 643 days 819 1.5% 9.6% (score = 2) 17.7%

3.8% (score = 1)Khorana et al. [54], 2010 Prospective **, 3 mos 30 – *** – 27%Moore et al. [13], 2011 Retrospective, cisplatin-based chemotherapy only, NA 932 13% 17.1% 28.2%Mandala et al. [41], 2011 Retrospective, phase I patients only, 2 months 1,415 1.5% 4.8% 12.9%George et al. [51], 2011 Subgroup analysis of SAVE-ONCO (placebo arm), 3.5 months 1,604 1.3% 3.5% 5.4%

NA = not available.* Pancreatic cancer patients are assigned a score of 2 based on site of cancer and therefore there were no patients in the low-risk category.** Included 4-weekly screening ultrasonography.*** Enrolled only high-risk patients.

thromboprophylaxis based on the known high risk of VTE in thehospitalized cancer population and extrapolation from the data inmedical patients [46,47]. Predictors of VTE in hospitalized cancerpatients have been identified and include site of cancer, use ofchemotherapy and comorbid conditions [2]. However, risk assess-ment is not routinely conducted nor recommended for hospitalizedcancer patients although it may be argued that such an approach isnecessary in the future.

6.2. Outpatient chemotherapy

Most VTE now occurs in the outpatient setting and major recentRCTs have focused on outpatient thromboprophylaxis for solidtumor patients receiving systemic therapy. In the context of thisreview of risk assessment, it is instructive to evaluate study designsfor these RCTs since they utilized very different approaches to riskassessment.

The two largest prophylaxis studies defined “high risk” eligibilitybased on 3 risk factors: primary site of cancer, advanced stage andconcomitant chemotherapy. The Prophylaxis of Thromboembolismduring Chemotherapy Trial (PROTECHT) study evaluated the efficacyof daily nadroparin, a LMWH, in locally advanced or metastaticlung, gastrointestinal, pancreatic, breast, ovarian, and head/neckcancers actively receiving chemotherapy [48]. Event rates werelow: 2% of the treatment group and 3.9% of the placebo groupdeveloped a TEE (one-sided 95% CI 0.303%, p = 0.02) with anon-significant increase in major bleeding. It should be noted thatin the contemporary era, breast and head/neck cancer patientshave not been typically considered high-risk for VTE and theirinclusion in this study may have reduced the overall event rate. Thesecond and largest study thus far in a broad cancer population wasSAVE-ONCO, a prospective, double-blind, multicenter study of 3,212patients with locally advanced or metastatic solid tumors (lung,pancreas, stomach, colorectal, bladder or ovary) randomized todaily subcutaneous semuloparin (a novel ultra-LMWH) or placebo[49]. Patients receiving prophylactic semuloparin had 64 % relativerisk reduction of VTE (hazard ratio: 0.36; 95% CI [0.21, 0.60]; p <

0.0001) but a lower absolute risk reduction (1.2 vs. 3.4%). There wasno significant increase in major bleeding. Semuloparin awaits FDAapproval for the indication of preventing VTE in cancer patientsreceiving chemotherapy and is not currently available.

Two other RCTs focused on a much narrower but also high-riskpopulation: pancreatic cancer. In the CONKO-004 study (currentlyonly available in abstract form), VTE occurred in 5.0% (8 of 160) ofpatients randomized to enoxaparin (1 mg/kg daily for 3 months,then 40 mg daily) versus 14.5% (22 of 152) in the observationarm (p < 0.01) [50]. In the FRAGEM study, a phase II randomizedtrial, patients were assigned to full therapeutic doses of dalteparinversus observation. All-type VTE during the dalteparin treatment

period (<100 days from randomization) was reduced from 23% to3.4% (p = 0.002), an 85% risk reduction. All-type VTE throughoutthe whole follow-up period was also reduced from 28% to 12%(p = 0.039), a 58% risk reduction. Lethal VTE (at <100 days) wasseen only in the control arm, 8.3% versus 0% (p = 0.057), RR =0.092, 95% CI (0.005–1.635) but overall survival was no differentbetween the two arms. These latter studies show that in high-riskpatients, extremely high event rates of VTE occur and can safelybe reduced but their findings apply only to a small niche of thecancer population (advanced pancreatic cancer patients receivingchemotherapy).

Which, then, of these two approaches to risk assessment is“best”? The former is more broad-based but leads to a highnumber needed to treat (44, in SAVE-ONCO). The latter approachis narrower but its applicability to other cancer populations and itsability to meaningfully reduce the public health burden of VTE inthis setting is limited. Subgroup analyses of PROTECHT and SAVE-ONCO suggest that the risk assessment tool discussed previouslymay be a useful compromise. In SAVE-ONCO, when the risk scorewas applied (Table 4), 550 (17.4%) of patients enrolled were definedto be at high risk of VTE, 1,998 (63.2%) were at moderate risk, and614 (19.4%) were at low risk (VTE risk score of ≥3, 1–2, or 0 points,respectively) [51]. Risk reduction was indeed greater in the high-risk subgroup (5.4% in the placebo arm vs 1.4% in the semuloparinarm, for score ≥3 [HR = 0.27] compared to the low-risk subgroup(1.3% vs. 1% respectively for score = 0 [HR = 0.71]).

Currently ongoing research efforts are focused on outpatientprophylaxis of ambulatory cancer patients based on novel ap-proaches to risk assessment. The University of Rochester and DukeUniversity are conducting an NIH-sponsored prospective studybased on the previously discussed risk score in which canceroutpatients at high risk for VTE (risk score ≥3) are randomizedto observation or dalteparin for 12 weeks (Clinicaltrials.gov NC-T00876915). In a biomarker-based approach, the MicroTEC studyis investigating enoxaparin in patients with pancreatic, lung andcolorectal cancer with elevated plasma TF microparticles (Clinical-trials.gov NCT00908960) and is expected to be reported in 2012.These data, along with ongoing biomarker studies and risk factorregistries, will further clarify the best approach to risk assessmentfor VTE in the cancer population. For now, however, targetedthromboprophylaxis is consistent with current therapeutic “per-sonalized medicine” paradigms in oncology, provides an optimalrisk-benefit ratio for individual patients, and remains the best op-portunity to reduce the burden of VTE and its consequences forpatients with cancer.

S14 A.A. Khorana / Thrombosis Research 129, Supplement 1 (2012) S10–S15

Acknowledgements

Dr. Khorana is supported by grants from the National CancerInstitute (K23 CA120587), the National Heart, Lung and BloodInstitute (R01HL095109) and the V Foundation

Conflict of interest statement

Dr. Khorana has received research support from Pharmacyclics,Leo Pharma, sanofi-aventis and Eisai; consulting fees and hono-raria from Roche/Genentech, Bristol Myers Squibb, sanofi-aventis,Johnson and Johnson, Isis Pharmaceuticals, Boehringer-Ingelheim,Daiichi-Sankyo and Leo Pharma.

References

[1] Heit JA, O’Fallon WM, Petterson TM, Lohse CM, Silverstein MD, Mohr DN, etal. Relative impact of risk factors for deep vein thrombosis and pulmonaryembolism: a population-based study. Arch Intern Med 2002;162:1245–8.

[2] Khorana AA, Francis CW, Culakova E, Kuderer NM, Lyman GH. Frequency, riskfactors, and trends for venous thromboembolism among hospitalized cancerpatients. Cancer 2007;110:2339–46.

[3] Khorana AA, Connolly GC. Assessing risk of venous thromboembolism in thepatient with cancer. J Clin Oncol 2009;27:4839–47.

[4] Blom JW, Doggen CJ, Osanto S, Rosendaal FR. Malignancies, prothromboticmutations, and the risk of venous thrombosis. Jama 2005;293:715–22.

[5] Francis CW. Clinical practice. Prophylaxis for thromboembolism in hospital-ized medical patients. N Engl J Med 2007;356:1438–44.

[6] Maraveyas A, Waters J, Roy R, Fyfe D, Propper D, Lofts F, et al. Gemcitabineversus gemcitabine plus dalteparin thromboprophylaxis in pancreatic cancer.Eur J Cancer 2011.

[7] Menapace LA, Peterson DR, Berry A, Sousou T, Khorana AA. Symptomatic andincidental thromboembolism are both associated with mortality in pancreaticcancer. Thromb Haemost 106:371–8.

[8] O’Connell CL, Boswell WD, Duddalwar V, Caton A, Mark LS, Vigen C, et al.Unsuspected pulmonary emboli in cancer patients: clinical correlates andrelevance. J Clin Oncol 2006;24:4928–32.

[9] Mandala M, Barni S, Floriani I, Isa L, Fornarini G, Marangolo M, et al.Incidence and clinical implications of venous thromboembolism in advancedcolorectal cancer patients: the “GISCAD-alternating schedule” study findings.Eur J Cancer 2009;45:65–73.

[10] Reynolds MW, Shibata A, Zhao S, Jones N, Fahrbach K, Goodnough LT.Impact of clinical trial design and execution-related factors on incidence ofthromboembolic events in cancer patients: a systematic review and meta-analysis. Curr Med Res Opin 2008;24:497–505.

[11] Khorana AA, Fine RL. Pancreatic cancer and thromboembolic disease. LancetOncol 2004;5:655–63.

[12] Chew HK, Davies AM, Wun T, Harvey D, Zhou H, White RH. The incidenceof venous thromboembolism among patients with primary lung cancer. JThromb Haemost 2008;6:601–8.

[13] Moore RA, Adel N, Riedel E, Bhutani M, Feldman DR, Tabbara NE, et al. Highincidence of thromboembolic events in patients treated with Cisplatin-basedchemotherapy: a large retrospective analysis. J Clin Oncol 29:3466–73.

[14] Khorana AA, Kuderer NM, Culakova E, Lyman GH, Francis CW. Developmentand validation of a predictive model for chemotherapy-associated thrombosis.Blood 2008;111:4902–7.

[15] Murray J. Myeloproliferative disorders. Clin Med 2005;5:328–32.[16] Khorana AA, Francis CW, Culakova E, Fisher RI, Kuderer NM, Lyman GH.

Thromboembolism in hospitalized neutropenic cancer patients. J Clin Oncol2006;24:484–90.

[17] Samama MM. An epidemiologic study of risk factors for deep vein thrombosisin medical outpatients: the Sirius study. Arch Intern Med 2000;160:3415–20.

[18] Heit JA, Silverstein MD, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ,3rd. Risk factors for deep vein thrombosis and pulmonary embolism: apopulation-based case-control study. Arch Intern Med 2000;160:809–15.

[19] Blom JW, Vanderschoot JP, Oostindier MJ, Osanto S, van der Meer FJ,Rosendaal FR. Incidence of venous thrombosis in a large cohort of 66,329 can-cer patients: results of a record linkage study. J Thromb Haemost 2006;4:529–35.

[20] Starling N, Rao S, Cunningham D, Iveson T, Nicolson M, Coxon F, et al. Throm-boembolism in patients with advanced gastroesophageal cancer treated withanthracycline, platinum, and fluoropyrimidine combination chemotherapy: areport from the UK National Cancer Research Institute Upper GastrointestinalClinical Studies Group. J Clin Oncol 2009;27:3786–93.

[21] Scappaticci FA, Skillings JR, Holden SN, Gerber HP, Miller K, KabbinavarF, et al. Arterial thromboembolic events in patients with metastatic car-

cinoma treated with chemotherapy and bevacizumab. J Natl Cancer Inst2007;99:1232–9.

[22] Nalluri SR, Chu D, Keresztes R, Zhu X, Wu S. Risk of venous thromboembolismwith the angiogenesis inhibitor bevacizumab in cancer patients: a meta-analysis. JAMA 2008;300:2277–85.

[23] Choueiri TK, Schutz FA, Je Y, Rosenberg JE, Bellmunt J. Risk of arterialthromboembolic events with sunitinib and sorafenib: a systematic review andmeta-analysis of clinical trials. J Clin Oncol 28:2280–5.

[24] Akl EA, Karmath G, Yosuico V, Kim SY, Barba M, Sperati F, et al. Antico-agulation for thrombosis prophylaxis in cancer patients with central venouscatheters. Cochrane Database Syst Rev 2007:CD006468.

[25] Bennett CL, Silver SM, Djulbegovic B, Samaras AT, Blau CA, Gleason KJ,et al. Venous thromboembolism and mortality associated with recombinanterythropoietin and darbepoetin administration for the treatment of cancer-associated anemia. JAMA 2008;299:914–24.

[26] Agnelli G, Bolis G, Capussotti L, Scarpa RM, Tonelli F, Bonizzoni E, et al. Aclinical outcome-based prospective study on venous thromboembolism aftercancer surgery: the @RISTOS project. Ann Surg 2006;243:89–95.

[27] Kroger K, Weiland D, Ose C, Neumann N, Weiss S, Hirsch C, et al. Riskfactors for venous thromboembolic events in cancer patients. Ann Oncol2006;17:297–303.

[28] Khorana AA, Francis CW, Culakova E, Lyman GH. Risk factors forchemotherapy-associated venous thromboembolism in a prospective observa-tional study. Cancer 2005;104:2822–9.

[29] Simanek R, Vormittag R, Ay C, Alguel G, Dunkler D, Schwarzinger I, et al. Highplatelet count associated with venous thromboembolism in cancer patients:results from the Vienna Cancer and Thrombosis Study (CATS). J ThrombHaemost 2010;8:114–20.

[30] Connolly GC, Khorana AA, Kuderer NM, Culakova E, Francis CW, Lyman GH.Leukocytosis, thrombosis and early mortality in cancer patients initiatingchemotherapy. Thromb Res 2010;126:113–8.

[31] Stender MT, Frokjaer JB, Larsen TB, Lundbye-Christensen S, Thorlacius-UssingO. Preoperative plasma D-dimer is a predictor of postoperative deep venousthrombosis in colorectal cancer patients: a clinical, prospective cohort studywith one-year follow-up. Dis Colon Rectum 2009;52:446–51.

[32] Ay C, Vormittag R, Dunkler D, Simanek R, Chiriac AL, Drach J, et al. D-dimerand prothrombin fragment 1 + 2 predict venous thromboembolism in patientswith cancer: results from the Vienna Cancer and Thrombosis Study. J ClinOncol 2009;27:4124–9.

[33] Kasthuri RS, Taubman MB, Mackman N. Role of tissue factor in cancer. J ClinOncol 2009;27:4834–8.

[34] Zwicker JI, Liebman HA, Neuberg D, Lacroix R, Bauer KA, Furie BC, et al.Tumor-derived tissue factor-bearing microparticles are associated with venousthromboembolic events in malignancy. Clin Cancer Res 2009;15:6830–40.

[35] Khorana AA, Ahrendt SA, Ryan CK, Francis CW, Hruban RH, Hu YC, et al.Tissue factor expression, angiogenesis, and thrombosis in pancreatic cancer.Clin Cancer Res 2007;13:2870–5.

[36] Han LY, Landen CN, Jr., Kamat AA, Lopez A, Bender DP, Mueller P, et al.Preoperative serum tissue factor levels are an independent prognostic factorin patients with ovarian carcinoma. J Clin Oncol 2006;24:755–61.

[37] Khorana AA, Francis CW, Menzies KE, Wang JG, Hyrien O, Hathcock J, etal. Plasma tissue factor may be predictive of venous thromboembolism inpancreatic cancer. J Thromb Haemost 2008;6:1983–5.

[38] Thaler J, Ay C, Weinstabl H, Dunkler D, Simanek R, Vormittag R, et al.Circulating procoagulant microparticles in cancer patients. Ann Hematol2011;90(4):447–53.

[39] Ay C, Mackman N, Key NS, Kaider A, Thaler J, Marosi C, et al. Microparticle-associated tissue factor activity and occurrence of venous thromboembolismin brain, gastrointestinal and pancreatic cancer. J Thromb Haemost 2011;O-MO-133.

[40] Ay C, Dunkler D, Marosi C, Chiriac AL, Vormittag R, Simanek R, et al.Prediction of venous thromboembolism in cancer patients. Blood 116:5377–82.

[41] Mandala M, Clerici M, Corradino I, Vitalini C, Colombini S, Torri V, et al.Incidence, risk factors and clinical implications of venous thromboembolismin cancer patients treated within the context of phase I studies: the “SENDOexperience”. Ann Oncol 2011.

[42] Palumbo A, Rajkumar SV, Dimopoulos MA, Richardson PG, San Miguel J,Barlogie B, et al. Prevention of thalidomide- and lenalidomide-associatedthrombosis in myeloma. Leukemia 2008;22:414–23.

[43] Samama MM, Cohen AT, Darmon J-Y, Desjardins L, Eldor A, Janbon C,et al. A Comparison of Enoxaparin with Placebo for the Prevention ofVenous Thromboembolism in Acutely ILL Medical Patients. N Engl J Med1999;341(11):793–800.

[44] Leizorovicz A, Cohen AT, Turpie AG, Olsson CG, Vaitkus PT, Goldhaber SZ. Ran-domized, placebo-controlled trial of dalteparin for the prevention of venousthromboembolism in acutely ill medical patients. Circulation 2004;110:874–9.

[45] Cohen AT, Davidson BL, Gallus AS, Lassen MR, Prins MH, Tomkowski W, et al.Efficacy and safety of fondaparinux for the prevention of venous thromboem-bolism in older acute medical patients: randomised placebo controlled trial.BMJ 2006;332:325–9.

A.A. Khorana / Thrombosis Research 129, Supplement 1 (2012) S10–S15 S15

[46] Lyman GH, Khorana AA, Falanga A, Clarke-Pearson D, Flowers C, JahanzebM, et al. American Society of Clinical Oncology Guideline: Recommendationsfor Venous Thromboembolism Prophylaxis and Treatment in Patients WithCancer. J Clin Oncol 2007.

[47] Wagman LD, Baird MF, Bennett CL, Bockenstedt PL, Cataland SR, Fanikos J,et al. Venous thromboembolic disease. NCCN. Clinical practice guidelines inoncology. J Natl Compr Canc Netw 2008;6:716–53.

[48] Agnelli G, Gussoni G, Bianchini C, Verso M, Mandala M, Cavanna L, etal. Nadroparin for the prevention of thromboembolic events in ambulatorypatients with metastatic or locally advanced solid cancer receiving chemo-therapy: a randomised, placebo-controlled, double-blind study. Lancet Oncol2009;10:943–9.

[49] Agnelli G, George DJ, Fisher W, Kakkar AK, Lassen MR, Mismetti P, Mouret P,Chaudhari U, Turpie AG. The ultra-low molecular weight heparin (ULMWH)semuloparin for prevention of venous thromboembolism (VTE) in patientswith cancer receiving chemotherapy: SAVE ONCO study. J Clin Oncol, 2011.ASCO Annual Meeting 2011;29:Abstract No. LBA9014.

[50] Riess UP, G. Deutschinoff, B. Opitz, M. Stauch, P. Reitzig, S. Hahnfeld, A. Hilbig,J. Stieler and H. Oettle A prospective, randomized trial of chemotherapy with

or without the low molecular weight heparin (LMWH) enoxaparin in patients(pts) with advanced pancreatic cancer (APC): Results of the CONKO 004 trial.J Clin Oncol 2009. ASCO Annual Meeting Proceedings (Post-Meeting Edition).2009;27.

[51] George D, Agnelli G, Fisher W, AK K, Lassen MR, Mismetti P, et al. VenousThromboembolism (VTE) Prevention with Semuloparin in Cancer PatientsInitiating Chemotherapy: Benefit–Risk Assessment by VTE Risk in SAVE-ONCO.Blood: Proceedings of Annual Meeting of American Society of Hematology2011. 2011;Abstract 206.

[52] Kearney JC, Rossi S, Glinert K, Henry D. Venous thromboembolism (VTE) andsurvival in a cancer chemotherapy outpatient clinic: a retrospective chartvalidation of a VTE prediction model. Blood 2009;Abstract 2503.

[53] Price LH, Nguyen MB, Picozzi VJ, Kozarek RA. Portal vein thrombosis inpancreatic cancer: natural history, risk factors and implications for patientmanagement. 2010 Gastrointestinal Cancer Symposium Program and Proceed-ings 2010;Abstract 143.

[54] Khorana AA, Herman K, Rubens D, Francis CW. A Predictive Risk Score forCancer-Associated Thrombosis: Role of Screening in A Prospective Study.Blood 2010;Abstract 3173.