Annals of Internal Medicine

Review

Systematic Review: The Trousseau Syndrome Revisited: Should We Screen Extensively for Cancer in Patients with Venous Thromboembolism?
Marc Carrier, MD; Gregoire Le Gal, MD, PhD; Philip S. Wells, MD, MSc; Dean Fergusson, PhD; Tim Ramsay, PhD; and Marc A. Rodger, MD, MSc

Background: Identifying previously undiagnosed cancer in patients with newly diagnosed venous thromboembolism (VTE) is important. Screening for malignant conditions can potentially diagnose more cases of cancer and at earlier stages, thereby preventing cancer-associated morbidity and perhaps mortality. Purpose: To summarize the period prevalence of previously undiagnosed cancer at baseline (within 1 month of VTE diagnosis), 6 months, and 12 months after VTE diagnosis and to quantify the additional value of an extensive cancer screening strategy (limited screening plus imaging techniques or tumor marker measurement) at baseline compared with more limited screening (history, physical examination, and simple widely available tests) at baseline. Data Sources: MEDLINE, EMBASE, the Cochrane Register of Controlled Trials, and Evidence-Based Medicine Reviews. Study Selection: A total of 36 studies that reported the prevalence of undiagnosed cancer at baseline, 6 months, and 12 months were selected. Fourteen articles and 1 abstract also met inclusion criteria for the assessment of extensive versus limited cancer screening. Data Extraction: Two reviewers independently extracted data onto standardized forms.

Data Synthesis: The period prevalence of previously undiagnosed cancer in patients with unprovoked VTE was 6.1% (95% CI, 5.0% to 7.1%) at baseline and 10.0% (CI, 8.6% to 11.3%) from baseline to 12 months. An extensive screening strategy using computed tomography of the abdomen and pelvis statistically significantly increased the proportion of previously undiagnosed cancer detected from 49.4% (CI, 40.2% to 58.5%) (with limited screening alone) to 69.7% (CI, 61.1% to 77.8%) in patients with unprovoked VTE. Limitation: The investigators could not determine complication rates, cost-effectiveness, and difference in morbidity and mortality associated with extensive screening strategies. Conclusion: Previously undiagnosed cancer is frequent in patients with unprovoked VTE. Many cases of previously undiagnosed cancer are missed by screening. An extensive cancer screening strategy detects more malignant conditions than does a limited screening strategy.

Ann Intern Med. 2008;149:323-333. For author affiliations, see end of text.

www.annals.org

eep venous thrombosis and pulmonary embolism, collectively termed venous thromboembolism (VTE), can be the rst manifestation of cancer (1 4). Armand Trousseau rst described this association in 1865: The diagnosis in a patient with gastric pain and leg or arm phlegmasia alba dolens is the presence of cancer (5). Two years later, on 1 January 1867, after suffering for weeks from abdominal pain, Trousseau declared to one of his students: Im lost. A thrombosis developed overnight, which doesnt leave me any doubt of the nature of my disease (5). Trousseau died of gastric cancer shortly thereafter. Identifying previously undiagnosed cancer in patients with VTE may be important for several reasons: It may be at a curable stage, or if detected later, may no longer be curable; treating cancer earlier may prevent cancer-associated morbidity (for example, pathologic fractures and compression syndromes); and cancer-associated VTE is now initially treated with low-molecular-weight heparin for 6 months rather than vitamin K antagonists (6), which can reduce the heightened risk for recurrent VTE with cancerassociated thrombosis (7) and potentially increase survival in patients without metastatic disease (8, 9). Although retrospective studies suggested that limited cancer screening, including a careful medical history, physical examination, and basic blood work, was adequate to

detect most cases of undiagnosed cancer in patients with VTE (10, 11), 2 recent prospective studies suggest that a more extensive screening strategy involving limited screening plus imaging studies or tumor marker measurement can increase the rate of cancer detection (12, 13). However, extensive screening may cause anxiety, may lead to test complications and unnecessary further testing for falsepositive results, and is invasive and expensive. To counsel patients with VTE on the risks and benets of cancer screening, clinicians require estimates of the absolute risk for previously undiagnosed cancer in patients and patient subgroups (provoked and unprovoked VTE),

See also: Print Editors Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 Editorial comment. . . . . . . . . . . . . . . . . . . . . . . . . . 350 Web-Only Appendix Tables CME quiz Conversion of graphics into slides Audio summary
2008 American College of Physicians 323

Review
Context

Cancer Screening in Patients with Venous Thromboembolism

Study Selection
Screening and surveillance for occult cancer in patients with unprovoked venous thromboembolism (VTE) is poorly understood.

Contribution
The authors systematically reviewed studies of occult cancer in patients with newly diagnosed unprovoked VTE. In 36 studies, the combined prevalence of occult cancer was 6.1% at baseline and 10.0% at 12 months after VTE diagnosis. The proportion of occult cancer diagnosed at baseline was 49.4% with limited screening and 69.7% with extensive screening.

Caution
The report did not describe the outcomes of newly diagnosed cancer associated with VTE.

Implication
Because only two thirds of cases of occult cancer are diagnosed at baseline, patients with unprovoked VTE need ongoing surveillance. The Editors

Using a structured question format to aid our literature search strategy, we reviewed abstracts of observational or randomized, controlled trials (RCTs) that reported the prevalence or incidence of previously undiagnosed cancer after VTE. To determine the period prevalence of previously undiagnosed cancer at baseline and during different follow-up intervals, we reviewed potentially relevant studies that satised all of the following criteria: study of adult patients with newly diagnosed VTE (deep venous thrombosis or pulmonary embolism); study that reported prevalence or incidence of cancer detection at diagnosis and at 6 or 12 months; and observational study or RCT. To assess the efcacy of limited and extensive cancer screening, we reviewed potentially relevant studies that satised all of the following criteria: study of adult patients with newly diagnosed VTE (deep venous thrombosis or pulmonary embolism); patients were screened for cancer by using history, physical examination, basic blood work, chest radiography (limited cancer screening), and serum tumor markers or abdominal ultrasonography or computed tomography (extensive cancer screening); observational study or RCT; and study reported 1 or more primary or secondary outcome measures.
Outcome Measures

estimates of the additional cases of cancer detected by extensive screening strategies compared with more limited screening strategies, and estimates of the potential additional risks and benets (morbidity and mortality) of an extensive screening strategy. To address these knowledge gaps, we performed a systematic review of the literature to summarize the period prevalence of previously undiagnosed cancer at baseline and during different intervals of follow-up, and the additional yield and complications of extensive screening strategies.

METHODS
Data Sources and Searches

We conducted a systematic literature search strategy to identify potential trials on MEDLINE (1950 to the rst week of November 2007), EMBASE (1980 to week 45 of 2007), the Cochrane Register of Controlled Trials (4th quarter, 2007) and Evidence-Based Medicine Reviews (4th quarter, 2007) by using the Ovid interface. We also sought publications through a hand-search of personal and potentially relevant journals and of the International Society on Thrombosis and Haemostasis conference proceedings for the past 2 years (2005 to 2007). Our systematic search strategy is documented in Appendix Table 1 (available at www.annals.org). We restricted the search to human trials. We made no restrictions on language, year of publication, or type of publication. We also searched the references of included studies and narrative reviews for potential studies.
324 2 September 2008 Annals of Internal Medicine Volume 149 Number 5

The primary outcome measure was detection of previously undiagnosed cancer. We dened previously undiagnosed cancer as any new case of cancer detected after the index VTE diagnosis until the end of follow-up in patients with no known history of cancer. We dened a prevalent case of previously undiagnosed cancer at baseline as a case of cancer detected at, or shortly after ( 1 month), the index VTE diagnosis in patients with no known history of cancer. We dened newly diagnosed cancer after baseline as a new case of cancer diagnosed between the index visit of VTE ( 1 month) and the end of follow-up. We dened limited cancer screening as a combination of medical history, physical examination, routine laboratory blood tests, and chest radiography. We dened extensive cancer screening as a combination of limited screening and 1 of the following: ultrasonography of the abdomen or pelvis, computed tomography of the abdomen or pelvis, or tumor markers. We followed the denitions of provoked VTE provided in the original studies. We dened unprovoked VTE as VTE with no predisposing factor. We excluded patients with a previous diagnosis of cancer. Secondary outcome measures included proportion of early-stage cancer, cancer-related mortality, and screening complications. We classied early-stage cancer by using the tumor/node/metastasis classication as T1-2N0M0. We dened screening complication as any complication arising directly from the screening (for example, biopsy) or from any other subsequent investigations to conrm or refute a malignant condition.
www.annals.org

Cancer Screening in Patients with Venous Thromboembolism

Review

Data Extraction and Quality Assessment

Two reviewers independently applied the 2 sets of inclusion criteria to the identied articles from the initial search strategy. They discussed articles for potential full review and independently extracted baseline characteristics, number of cases of previously undiagnosed cancer, time of diagnosis (baseline, 2 to 6 months, or 7 to 12 months), type and stage of cancer, and diagnostic test used for diagnosis. The 2 reviewers extracted data for all patients with any, unprovoked, or provoked VTE. A third reviewer noted and adjudicated discrepancies. A priori, we planned to evaluate the methodological quality of observational studies by using the validated NewcastleOttawa Quality Assessment scales (14). The NewcastleOttawa quality score was designed to assess the quality of nonrandomized studies in meta-analyses (14). The scale assesses cohort and casecontrol studies by using 3 broad perspectives: the selection of the study groups; the comparability of the groups; and the ascertainment of the outcome of interest or the exposure for cohort studies and casecontrol studies, respectively. We also extracted data identied as being important markers of quality in case series and cohort studies (explicit inclusion criteria, length of follow-up, and outcomes assessed by using objective criteria) (15). Post hoc, we also used the NewcastleOttawa scale to assess the RCT quality because there was only 1 RCT, and we used the 2 groups of the RCT as 2 cohorts without comparing outcomes between them. We further assessed quality of RCTs by examining allocation concealment (16). For all eligible studies, 2 reviewers independently assessed trial or study quality and extracted the data by using a standardized data abstraction form. Likewise, a third reviewer documented and adjudicated any discrepancies.
Data Synthesis and Statistical Analysis

patients with VTE and in subgroups based on the cause of VTE (provoked and unprovoked). Because cancer screening was performed shortly after the index VTE, we used a cross-sectional design to assess whether extensive screening provided any additional benets in cancer detection. We estimated the pooled probability of identifying previously undiagnosed cancer by using a limited or extensive screening strategy. We also determined the proportions of previously undiagnosed cancer detected by using different diagnostic tests (ultrasonography, computed tomography, and tumor markers) within the different extensive screening strategies. The proportions represent the number of cases of previously undiagnosed cancer detected by the specied screening tool over the total number of cancer cases (prevalent and incident) detected in the population at the end of follow-up. We calculated the CIs by using averaged, inverse varianceweighted estimates from each study.
Role of the Funding Source

The Canadian Institute of Health Research provided funding for the study. The funding source had no role in the design, conduct, or analysis of the study or in the decision to submit the manuscript for publication. Dr. Carrier had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

RESULTS
Our literature search identied 939 citations, and we deemed 36 articles (1 abstract, 35 full articles) potentially eligible (Figure) (10 13, 17 48). We excluded 2 studies that were not written in English, French, or Spanish because not enough information could be extracted from the English language abstract (47, 48).
Period Prevalences

We dened period prevalence as the proportion of the VTE population with previously undiagnosed cancer detected over a specic period of time. To estimate the period prevalence at baseline and after different intervals of follow-up, we extracted data from the individual studies (including the 2 groups of the RCT [13]) to form a cohort of patients followed over time. We dened the cohort as a group of patients without previously known cancer at the time of the VTE diagnosis. The period prevalence at baseline includes all prevalent cases of previously undiagnosed cancer from the time of VTE diagnosis to 1 month. The period prevalences, between 2 to 6 months and between 7 to 12 months, include all cases of newly diagnosed cancer during follow-up. The period prevalence from baseline to 6 months and baseline to 12 months combines all cases of previously undiagnosed cancer over the specic time periods. We calculated the 95% CI for each period prevalence by using averaged, inverse varianceweighted estimates from each study. We determined period prevalences for all
www.annals.org

We included 9516 patients with VTE (3286 unprovoked, 3297 provoked, and 2933 unspecied) from 34 studies in the assessments of period prevalence of newly diagnosed cancer (10 13, 17 46). Table 1 shows the period prevalence at baseline and during different intervals of follow-up of previously undiagnosed cancer after VTE. The period prevalence of previously undiagnosed cancer detection from baseline to 12 months was 6.3% (95% CI, 5.6% to 6.9%) in all patients with VTE. The period prevalence of previously undiagnosed cancer between baseline and 12 months was higher in patients with unprovoked VTE (10.0% [CI, 8.6% to 11.3%]) than in those with provoked VTE (2.6% [CI, 1.6% to 3.6%]). The period prevalences of newly diagnosed cancer were similar within the rst and second 6 months of follow-up after presentation with VTE (Table 1).
Extensive versus Limited Cancer Screening

Of the 34 articles used for period prevalence assessment, 14 articles and 1 abstract also met the second set of
2 September 2008 Annals of Internal Medicine Volume 149 Number 5 325

Review

Cancer Screening in Patients with Venous Thromboembolism

Figure. Study flow diagram.

Total articles found (n = 1320) MEDLINE (n = 391) EMBASE (n = 839) Cochrane (n = 90) Duplicates removed (n = 381) Records screened (n = 939)

Records excluded (n = 903)

Article records assessed for eligibility (n = 36) Articles excluded for not being in English, French, or Spanish (n = 2)

We included 4378 patients from all 15 studies. Sample size ranged from 21 to 864 patients. Baseline characteristics were similar among the trials (Table 2). Thirteen studies were cohort studies; 4 were retrospective and 9 were prospective. One study prospectively compared an exposed cohort with a nonexposed cohort (different centers) (45), and 1 was an RCT (13). We assessed quality by using the NewcastleOttawa scale (Appendix Table 2, available at www.annals.org) (14). Thirteen cohort studies did not have an exposed (cancer screening) and nonexposed cohort (no screening); therefore, we could not assess comparability of cohorts (Appendix Table 2, available at www.annals .org). We also evaluated the quality of the RCT by using the same quality scale because the 2 groups of the RCT were used as cohorts without comparing outcomes between them. The RCT did not report allocation concealment (13). (text continues on page 329)

Articles included to assess cumulative incidence (n = 34)

Table 1. Period Prevalence of Previously Undiagnosed Cancer in Patients with VTE

Time Point and Type of VTE (Reference) Articles excluded for not assessing extensive screening strategies (n = 19) Baseline (<1 mo) Overall (1113, 1721, 2328, 3132, 3446) Unprovoked VTE (13, 1718, 21, 2324, 2628, 31, 3536, 3840, 4344, 46) Provoked VTE (1718, 21, 2324, 27, 31, 35, 3840, 44) >1 to <6 mo Overall VTE (1113, 17, 1921, 2325, 2728, 3536, 38, 4041, 4344, 46) Unprovoked VTE (17, 21, 2324, 2728, 3536, 38, 40, 4344, 46) Provoked VTE (17, 21, 2324, 27, 35, 38, 40, 44) Baseline to 6 mo (0 to <6 mo) Overall (1113, 17, 1921, 2325, 2728, 3536, 38, 4041, 4344, 46) Unprovoked VTE (17, 21, 2324, 2728, 3536, 38, 40, 4344, 46) Provoked VTE (17, 21, 2324, 27, 35, 38, 40, 44) >6 to <12 mo Overall (1113, 17, 1921, 2325, 2728, 3536, 38, 4041, 4344, 46) Unprovoked (13, 17, 21, 2324, 2728, 3536, 38, 40, 4344, 46) Provoked (17, 21, 2324, 27, 35, 38, 40, 44) Baseline to 12 mo (0 to <12 mo) Overall (1113, 17, 2021, 2329, 3340, 4344, 46) Unprovoked (13, 17, 2021, 2324, 2628, 3536, 3840, 4344, 46) Provoked (17, 21, 2324, 27, 35, 38, 40, 44) Period Prevalence of Previously Undiagnosed Cancer (95% CI), % 4.1 (3.64.6) 6.1 (5.07.1) 1.9 (1.32.5)

Articles that met all inclusion criteria (n = 14) Abstract that met all inclusion criteria (n = 1)

1.1 (0.71.6) 2.0 (1.12.9) 0.4 (01.3)

inclusion criteria for the efcacy assessment of extensive versus limited cancer screening (Figure). There were no discrepancies between the 2 reviewers with regard to study eligibility (12, 13, 18, 35 46). Most studies dened limited screening as a combination of careful medical history, physical examination, laboratory blood testing, and chest radiography. Extensive screening included ultrasonography or computed tomography of the abdomen and pelvis or measurement of tumor markers (prostate-specic antigen, carcinoembryonic antigen, or cancer antigen-125). One study also included gastroscopy or barium swallow, colonoscopy or barium enema, mammography, and sputum cytology (13). Most studies did not report the performance (sensitivity and specicity) of the individual diagnostic tests (for example, computed tomography) used in screening. The limited and extensive screenings were completed within 2 to 4 weeks of the index VTE in most studies. No studies performed regular testing during follow-up. New cancer diagnoses were recorded if patients sought medical attention or by review of their medical charts.
326 2 September 2008 Annals of Internal Medicine Volume 149 Number 5

5.4 (4.86.0) 8.6 (7.210.0) 2.4 (1.43.4)

1.2 (0.71.6) 1.4 (0.62.2) 0.5 (01.1)

6.3 (5.66.9) 10.0 (8.611.3) 2.6 (1.63.6)

VTE

venous thromboembolism.
www.annals.org

Cancer Screening in Patients with Venous Thromboembolism

Review

Table 2. Characteristics of Included Studies

Author, Year (Reference) Purpose of Study Inclusion Criteria Limited Screening Strategy (Patients Tested) Extensive Screening Strategy (Patients Tested) Testing per Protocol Follow-up Method Length of Follow-up Patients Lost to Followup, n Outcome Assessment

Retrospective cohort studies Sannella and OConnor, 1991 (36)

Fahrig et al., 1998 (42)

Netzer et al., 1999 (43)

Assess CT as a cancer screening method in patients with unprovoked DVT Assess the incidence of cancer in patients with DVT Assess US abdo/ pelvis in patients with unprovoked DVT

Patients with DVT; only unprovoked DVT evaluated (n 21) Patients with DVT and no known cancer Consecutive patients with unprovoked DVT Patients with new DVT

Hx, PE, CBC, CXR (n 21)

CT abdo/pelvis (n 17)

NS

Chart review

8y

NS

Laboratory investigations, CXR (unknown) Other means during hospitalization (unknown)

US abdo/pelvis (unknown)

NS

NS

NS

NS

US abdo/pelvis (n 135)

No

Ronsdorf et al., 2003 (44)

Assess the prevalence of cancer in patients with provoked and unprovoked DVT

Hx, PE, CBC, electrolytes, ESR, creat, UA (unknown); CXR (n 479)

US abdo/pelvis (n 385)

Yes

Chart reviews, telephone interviews, autopsy reports Mailed questionnaires, chart reviews, death certificates

Mean, 30 mo

Histology

Mean, 33 mo

51

NS

Prospective cohort studies Monreal et al., 1991 (35)

Validate extensive cancer screening on all patients with DVT

Consecutive patients with DVT and no known cancer Patients with unprovoked DVT and no known cancer Consecutive patients with pulmonary embolism and no known cancer Patients with DVT

Hx, PE, CBC, LFTs, ESR, LDH, SPEP, CXR, blood smear (n 113)

Barrelier et al., 1992 (37)

Assess diagnostic testing to determine the cause of unprovoked VTE

Monreal et al., 1993 (38)

Validate extensive cancer screening on all patients with pulmonary embolism

Hx, PE, CBC, iron, ferritin, uric acid, SPEP, cholesterol, LFTs, ANA, hemostasis factors (unknown) Hx, PE, CBC, LFTs, ESR, LDH, SPEP, CXR, blood smear (n 78)

CEA (113); US abdo/pelvis (n 113); CT abdo/pelvis (n 69); UGI endoscopy (n 83) US abdo/pelvis (unknown)

Yes

Clinic visits

Mean, 1215 mo

NS

Yes

NS

12 mo

NS

CEA (n 78); US abdo/pelvis (n 78); UGI endoscopy (n 55)

Yes

Clinic visits

Mean, 921 mo

11

NS

Pistorius et al., 1994 (39)

Assess diagnostic testing to determine the cause of DVT Assess the incidence of cancer in patients with DVT

Bastounis et al., 1996 (40)

Consecutive patients with first episode of DVT and no known cancer

Hx, PE, CBC, LFTs, ESR, CRP, fibrinogen, SPEP, CXR (n 53) Hx, PE, CBC, ESR, LFTs, creat, SPEP, UA, CXR (n 293)

US abdo/pelvis (n 53)

Yes

Clinic visits

1y

NS

CEA (n 293); CT abdo/pelvis (n 222)

Yes

Clinic visits

2y

NS

Continued on following page

www.annals.org

2 September 2008 Annals of Internal Medicine Volume 149 Number 5 327

Review

Cancer Screening in Patients with Venous Thromboembolism

Table 2Continued
Author, Year (Reference) Purpose of Study Inclusion Criteria Limited Screening Strategy (Patients Tested) Extensive Screening Strategy (Patients Tested) US abdo/pelvis (n 143) Testing per Protocol Follow-up Method Length of Follow-up Patients Lost to Followup, n 27 Outcome Assessment

Cailleux et al., 1997 (41)

Monreal et al., 1997 (18)

Assess US as a cancer screening method in hospitalized patients with VTE Assess cancer screening diagnostic tests in patients with VTE

Consecutive patients with VTE and no known cancer Consecutive patients with DVT and no known cancer

Hx, PE, basic blood work, CXR (unknown)

No

Clinic visit

3 to 6 mo

NS

Hx, PE, CBC, LFTs, ESR, LDH, SPEP, CXR, blood smear (n 674)

Engudanos et al., 2002 (46)

Assess tumor markers as a cancer screening method in patients with unprovoked DVT Assess cancer screening diagnostic tests in patients with VTE

Consecutive patients with unprovoked DVT

Hx, PE, laboratory assessment (n 48)

Monreal et al., 2004 (12)

van Doormaal Assess CT as a et al., cancer screen2007 (45)* ing in patients with unprovoked VTE Randomized, controlled trial Piccioli et al., 2004 (13)

Consecutive patients with new VTE and no known cancer Patients with unprovoked VTE

Hx, PE, CBC, LFTs, creat, ESR, SPEP, UA, CXR (n 864)

Hx, PE, basic laboratory examination, CXR (n 444)

CEA (n 674); PSA (unknown); US abdo (n 569); CT abdo/ pelvis (n 105) CEA, CA 19-9, SCC, NSE, CA 15-3, -FP, 2-microglobulin (n 48); PSA (n 33); CA 125 (n 15) CEA, US abdo/ pelvis (830); PSA (unknown); CA 125 (unknown) CT thorax, abdo/ pelvis (n 214)

Yes

Clinic visit

13 y

NS

NS

Yes

Clinic visits

Median, 10 mo

Histology (6 of 8 patients)

Yes

Clinic visits

12 mo

28

Histology

Yes

NS

Median, 12 mo

NS

Histology

Assess extensive cancer screening in patients with unprovoked VTE with normal limited screening

Consecutive patients with unprovoked VTE and no known cancer

Hx, PE, CBC, AST, ALT, ALP, calcium, UA, CXR (n 339)

US abdo/pelvis (n 114); CT abdo/pelvis (n 95); UGI endoscopy or barium swallow (n 84); colonoscopy, barium enema or sigmoidoscopy (n 84); CEA, -FP, CA 125 (n 119); hemoccult (n 110); sputum cytology (n 97); Pap smear (n 59); mammography (n 38); PSA (n 65); prostate US (n 65)

Yes

Clinic visits

24 mo

Histology

-FP -fetoprotein; abdo abdomen; ANA antinuclear antibodies; ALP alkaline phosphatase; ALT alanine aminotransferase; AST aspartate aminotransferase; CA cancer antigen; CBC complete blood count; CEA carcinoembryonic antigen; creat creatinine; CRP C-reactive protein; CT computed tomography; CXR chest radiography; DVT deep venous thrombosis; ESR erythrocyte sedimentation rate; Hx history; LDH lactate dehydrogenase; LFT liver function test; NS not specied; NSE neuron-specic enolase; Pap Papanicolaou; PE physical examination; PSA prostate-specic antigen; SPEP serum protein electrophoresis; UA urine analysis; US ultrasonography; SCC squamous-cell antigen; UGI upper gastrointestinal; VTE venous thromboembolism. * Prospective cohort study with concurrent controls. Randomized, controlled trial assessing clinical outcome in patients with unprovoked VTE in whom limited clinical examination did not reveal cancer (extensive screening vs. control participants).
328 2 September 2008 Annals of Internal Medicine Volume 149 Number 5 www.annals.org

Cancer Screening in Patients with Venous Thromboembolism

Review

Seven studies (713 patients) reported the rate of previously undiagnosed cancer detected in patients with unprovoked VTE (13, 16, 35, 36, 38, 43, 46). Five studies reported the tumor/node/metastasis staging of the cancer detected (12, 13, 35, 38, 44). One study reported overall and cancer-related mortality (13). Although sensitivity and subgroup analyses were planned, the small number of studies precluded some analyses. Table 3 shows the additional value of extensive screening to detect previously undiagnosed cancer. Extensive screening increased the proportion of previously undiagnosed cancer that was detected. Computed tomography of the abdomen and pelvis statistically signicantly increased the proportion of undiagnosed cancer detected from 47.6% (CI, 40.0% to 55.1%) with limited screening alone to 66.1% (CI, 59.0% to 73.2%) in all patients with VTE, and from 49.4% (CI, 40.2% to 58.5%) to 69.7% (CI, 61.1% to 77.8%) in the subgroup of patients with unprovoked VTE (Table 3). Although ultrasonography of the abdomen and pelvis, carcinoembryonic antigen screening, and prostate-specic antigen screening increased the frequency of cancer detection, their CIs overlap with those from the group with limited screening alone. Table 4 reports the number of prevalent cases of previously undiagnosed cancer detected by screening and cases of newly diagnosed cancer (that is, those missed by screening). Appendix Table 3 (available at www.annals.org) shows frequencies of the various types of cancer detected by screening. Four studies compared the rate of detection of earlystage, previously undiagnosed cancer (T1-2N0M0) between the limited and extensive screening strategies (12, 35, 38, 44). Two studies showed a benet of extensive screening in early cancer detection (12, 35), whereas the other 2 did not

(38, 44). A total of 123 cases of cancer were detected in these 4 studies. Extensive screening (pooled data across the different diagnostic tests) increased the number of cases of early-stage, previously undiagnosed cancer detected from 17 (13.8% [CI, 8.8% to 21.0%]) with limited screening only to 27 (22.0% [CI, 15.6% to 30.1%]) (12, 35, 38, 44). Only 1 study reported the rate of cancer-related mortality according to the screening strategy used (13). An absolute difference in cancer-related mortality of 1.9% (CI, 5.5% to 10.9%) was detected in favor of the extensive screening strategy. No studies reported screening complications.

DISCUSSION
Our systematic review demonstrates that previously undiagnosed cancer is frequent in patients with unprovoked VTE. The period prevalence of undiagnosed cancer is 6.1% (CI, 5.0% to 7.1%) at baseline and 10.0% (CI, 8.6% to 11.3%) from the time of VTE diagnosis to 12 months. Our review adds to the literature by pooling data from cohort studies and RCTs and by reporting pooled estimates of the period prevalence of previously undiagnosed cancer at baseline (prevalent cases) and during subsequent follow-up up to 1 year. To our knowledge, this systematic review is also the largest and most comprehensive assessment of the incremental value of an extensive screening strategy compared with a more limited one. Our review showed that use of an extensive screening strategy, in particular, computed tomography of the abdomen and pelvis, detects more malignant conditions than a limited screening strategy, without reported complications. However, because of the lack of detail in the studies, it is still unclear whether an increase in detection of new malignant

Table 3. Proportion of Previously Undiagnosed Cancer Cases Detected at Baseline by Limited or Extensive Screening
Screening Technique and Type of VTE (Reference) Proportion of Previously Undiagnosed Cancer Cases (95% CI), %* Limited Screening CT of abdomen/pelvis All VTE (13, 18, 35, 40, 45) Unprovoked VTE (13, 18, 35, 45) US of abdomen/pelvis All VTE (12, 13, 35, 37, 38, 41, 43, 44) Unprovoked VTE (13, 35, 38, 41, 43, 44) Tumor marker CEA All VTE (12, 18, 35, 38, 40) Unprovoked VTE (35, 38) Tumor marker PSA All VTE (12, 13, 18) Unprovoked VTE (13, 18) 47.6 (40.055.1) 49.4 (40.258.5) Extensive Screening 66.1 (59.073.2) 69.7 (61.177.8)

54.1 (47.460.7) 54.2 (45.562.9)

61.9 (55.468.5) 63.5 (54.972.1)

47.8 (39.556.0) 66.7 (28.9100)

53.2 (44.961.5) 83.3 (53.5100)

50.5 (42.758.3) 51.0 (40.062.0)

62.4 (54.770.1) 60.6 (49.671.7)

CEA carcinoembryonic antigen; CT computed tomography; PSA prostate-specic antigen; US ultrasonography; VTE venous thromboembolism. * The denominator represents the total cases of cancer (prevalent and incident) detected in the population at the completion of follow-up for the specied studies. Limited screening plus CT, US, or CEA or PSA measurement.
www.annals.org 2 September 2008 Annals of Internal Medicine Volume 149 Number 5 329

Review

Cancer Screening in Patients with Venous Thromboembolism

Table 4. Prevalent and Incident Cases of Previously Undiagnosed Cancer

Author, Year (Reference) Patients with VTE, n Total Cases of Previously Undiagnosed Cancer, n Prevalent Cases Diagnosed by Limited Screening Alone, n 8 4 Prevalent Cases Diagnosed by Extensive Screening, n* 11 7 Incident Cases Diagnosed during Follow-up, n Patients without New Diagnosis of Cancer, n

Monreal et al., 1991 (35) All Unprovoked VTE Sannella and OConnor, 1991 (36) All Unprovoked VTE Barrelier et al., 1992 (37) All Unprovoked VTE Monreal et al., 1993 (38) All Unprovoked VTE Pistorius et al., 1994 (39) All Unprovoked VTE Bastounis et al., 1996 (40) All Unprovoked VTE Cailleux et al., 1997 (41) All Unprovoked VTE Monreal et al., 1997 (18) All Unprovoked VTE Fahrig et al., 1998 (42) All Unprovoked VTE Netzer et al., 1999 (43) All Unprovoked VTE Engudanos et al., 2002 (46) All Unprovoked VTE Ronsdorf et al., 2003 (44) All Unprovoked VTE Monreal et al., 2004 (12) All Unprovoked VTE Piccioli et al., 2004 (13) All Unprovoked VTE van Doormaal et al., 2007 (45) All Unprovoked VTE Total All Unprovoked VTE

113 31

12 7

1 0

101 24

228 (only 21 followed) 21

11 11

3 3

10 10

1 1

217 (only 10 followed) 10

125 NA

3 NA

2 NA

3 NA

0 NA

122 NA

78 27

9 7

6 4

7 5

2 2

69 20

53 17

9 9

8 8

8 8

1 1

44 8

293 86

29 NA

13 NA

24 NA

7 NA

264 86

148 36

6 NA

5 NA

6 NA

0 NA

142 36

685 112

34 23

11 7

26 16

8 7

651 89

318 NA

24 NA

20 NA

24 NA

0 NA

294 NA

135 135

14 14

8 8

10 10

4 4

121 121

48 48

8 8

5 5

6 6

0 2

40 40

485 236

42 27

14 10

16 11

26 16

443 209

864 345

61 NA

34 NA

47 NA

14 NA

803 345

339 339

56 56

32 32

45 45

11 11

283 283

444 444

28 28

12 12

16 16

12 12

416 416

4356 1877

346/4356 190/1877

181/346 93/190

259/346 134/190

87/346 56/190

4010/4356 1687/1877

NA not applicable; VTE venous thromboembolism. * Pooled data across the different extensive screening strategies. Number of provoked VTE events diagnosed by limited and extensive screening not reported.
330 2 September 2008 Annals of Internal Medicine Volume 149 Number 5 www.annals.org

Cancer Screening in Patients with Venous Thromboembolism

Review

conditions will result in a statistically signicant change in the detection rate of early-stage, previously undiagnosed cases of cancer or a decrease in cancer-related morbidity, cancer-related mortality, or overall mortality. Furthermore, we could not assess important consequences of extensive screening, such as anxiety, complications (including radiation exposure), discomfort from extensive screening procedures, burden of additional tests for false-positive results, and cost-effectiveness. Our study showed that patients with VTE are at high risk for having previously undiagnosed cancer or developing cancer after VTE diagnosis. Patients with unprovoked VTE are at the highest risk. This nding is consistent with 2 large population registry studies (1, 10). One of the population registry studies showed that the highest risk for detection of malignant conditions occurred within the rst 6 months and decreased to baseline levels 12 months after presentation with VTE (10). However, prevalent and incident cases of previously undiagnosed cancer could not be separated. Our estimates showed that the period prevalence of newly diagnosed cancer does not substantially decrease after 6 months (that is, period prevalence is similar between 2 to 6 months and between 7 to 12 months) (Table 1). Most detected cases of previously undiagnosed cancer are present at baseline, and a more extensive screening strategy in the highest-risk patients can potentially increase the number of cancer cases detected. These previously undiagnosed cases may be important to diagnose for a variety of reasons. First, patients with cancer are at higher risk for recurrent VTE and bleeding despite adequate anticoagulation with vitamin K antagonists (49). The overall incidence of recurrent VTE in patients with cancer is 30.0 events per 100 patient-years, compared with 12.8 events per 100 patient-years in those without cancer, whereas the incidence of major bleeding is 15.7 and 8.6 events per 100 patient-years, respectively (49). Therefore, cancer-related VTE is best treated with at least 6 months of low-molecular-weight heparin to prevent recurrent VTE (6). Secondary prophylaxis with low-molecular-weight heparin was shown to decrease the rate of recurrent VTE by 52% at 6 months (7) and potentially increase survival in patients without metastatic disease (8) compared with vitamin K antagonists. Second, detecting undiagnosed cancer earlier might lead to an increase in the rate of curable cases and potentially result in increased survival. Population registry studies have suggested that approximately 40% of cases of previously undiagnosed cancer diagnosed during follow-up were already metastatic at the time of diagnosis (1, 2). Unfortunately, only 4 studies reported the frequency of detection of early-stage, previously undiagnosed cancer when extensive versus limited occult cancer screening was used (12, 35, 38, 44). Combining the prevalence of these 4 studies showed that an extensive screening strategy (pooled results from all diagnostic tests) detected more cases of early-stage, previously undiagnosed cancer than did the limited screening strategy. However,
www.annals.org

the few cases detected in each individual trial limit the interpretation of the pooled result. One study evaluated the impact of extensive occult cancer screening on cancer prognosis (13). The investigators found an absolute 1.9% (CI, 5.5% to 10.9%) reduction in cancer-related mortality, but the study was underpowered to nd a statistically signicant difference. Only 201 of the projected 1000 patients deemed required by the sample size calculation were enrolled in the study (13). Physicians favoring extensive screening and the use of a Zelen randomization procedure were major obstacles to successful completion of the study (50). Finally, detecting cancer earlier may prevent cancerassociated morbidity, potentially leading to an increase in quality of life. Further research is urgently required to evaluate the impact of detection of previously undiagnosed cancer on mortality and quality of life. Whereas some narrative reviews recommended extensive screening strategies for cancer detection in patients with unprovoked VTE (the group with highest risk for prevalent cancer) (51, 52), others have not (53, 54). Our systematic review showed that a more extensive screening strategy detects more cases of previously undiagnosed cancer. However, before recommending an extensive screening strategy to all patients with unprovoked VTE, we must consider some adverse effects, including morbidity and cost of extensive screening, risks of invasive diagnostic procedure to conrm cancer, incidental ndings and their associated morbidity, and psychological burden associated with new diagnosis of cancer or false-positive results. Furthermore, the components of an ideal extensive cancer screening strategy are still unknown. Our results favor computed tomography of the abdomen and pelvis, but the previously mentioned factors, including radiation exposure, need to be considered. Recent advances in diagnostic imaging tests (such as 3-dimensional computed tomography display and positron emission tomography) may help to improve the rate of detection of previously undiagnosed cancer in patients with VTE. Studies are currently under way to evaluate the impact of new diagnostic technologies in the detection of undiagnosed cancer. Our study has limitations. First, we could not calculate the sensitivity and specicity of both screening strategies. Ideally, we would have reported the sensitivities and specicities for each screening strategy and its component individual modalities; however, most studies did not report false-positive results (positive diagnostic test results without a nal diagnosis of cancer), making calculation of sensitivities and specicities impossible. Second, the cost of extensive screening can be substantial and we did not account for this in our analysis. Cost-effectiveness analysis should be performed alongside any future trial regarding this clinical question. Third, we pooled results of retrospective and prospective studies, which may have led to an overestimation of the overall effect. Fourth, although all studies included in our analysis excluded patients with previous cancer, the denition of provoked VTE, based on other
2 September 2008 Annals of Internal Medicine Volume 149 Number 5 331

Review

Cancer Screening in Patients with Venous Thromboembolism

predisposing factors, was heterogeneous among the studies. Some studies used less well-accepted denitions of provoked VTE by including weaker provoking factors, such as known thrombophilia, leg varices, previous VTE, congestive heart failure, and peripheral vascular disease. Individual patient data on predisposing factors of VTE and study outcomes were not reported in most studies, which made it difcult to apply a uniform denition of provoked and unprovoked VTE. This may have led to some patients with unprovoked VTE, as dened by more accepted definitions, being categorized in our study as patients with provoked VTE. However, this would have led to bias toward a null effect when comparing the period prevalence of newly diagnosed cancer in provoked VTE with that in patients with unprovoked VTE and probably led to lower estimates of period prevalence in patients with unprovoked VTE and higher estimates of period prevalence in patients with provoked VTE. Fifth, the proportions of previously undiagnosed cancer detected (Table 3) are biased to shortterm follow-up and might be different in longer-term follow-up because more cases of previously undiagnosed cancer may become evident. Sixth, we could not review all nonEnglish-language publications. Finally, we planned sensitivity and subgroup analyses, but the small number of studies limited the number of analyses and the condence in the effect estimates. In conclusion, we found that previously undiagnosed cancer is common in patients with unprovoked VTE and that cancer is more likely to be detected by an extensive screening strategy, including computed tomography of the abdomen and pelvis. However, many cases of previously undiagnosed cancer remain undetected after screening, and clinicians should monitor their patients closely for any signs of malignant conditions. What remains unknown is whether detection reduces morbidity, ameliorates quality of life, is cost-effective, or improves survival.
From the University of Ottawa and the Ottawa Health Research Institute, Ottawa, Ontario, Canada, and Brest University Hospital, Brest, France.
Grant Support: Dr. Carrier is a recipient of a Canadian Institutes of Health Research Fellowship. Dr. Fergusson is a recipient of a Canadian Institutes of Health Research New Investigator Research Award. Dr. Rodger is the recipient of a Career Scientist Award from the Heart and Stroke Foundation of Ontario. Dr. Wells is a recipient of a Canada Research Chair. Potential Financial Conflicts of Interest: Honoraria: P.S. Wells (Dade Behring, bioMerieux, Sano-Aventis, Leo Pharma, Organon). Requests for Single Reprints: Marc A. Rodger, MD, MSc, Clinical

References
1. Baron JA, Gridley G, Weiderpass E, Nyren O, Linet M. Venous thrombo embolism and cancer. Lancet. 1998;351:1077-80. [PMID: 9660575] 2. Srensen HT, Mellemkjaer L, Steffensen FH, Olsen JH, Nielsen GL. The risk of a diagnosis of cancer after primary deep venous thrombosis or pulmonary embolism. N Engl J Med. 1998;338:1169-73. [PMID: 9554856] 3. Lee AY, Levine MN. Venous thromboembolism and cancer: risks and outcomes. Circulation. 2003;107:I17-21. [PMID: 12814981] 4. White RH, Chew HK, Zhou H, Parikh-Patel A, Harris D, Harvey D, et al. Incidence of venous thromboembolism in the year before the diagnosis of cancer in 528,693 adults. Arch Intern Med. 2005;165:1782-7. [PMID: 16087828] 5. Trousseau A. Clinique medicale de lHotel-Dieu de Paris. Geneve-Paris-Brux ` elles: Alliance Culturelle du Livre; 1963. 6. Lyman GH, Khorana AA, Falanga A, Clarke-Pearson D, Flowers C, Jahanzeb M, et al. American Society of Clinical Oncology. American Society of Clinical Oncology guideline: recommendations for venous thromboembolism prophylaxis and treatment in patients with cancer. J Clin Oncol. 2007;25:5490505. [PMID: 17968019] 7. Lee AY, Levine MN, Baker RI, Bowden C, Kakkar AK, Prins M, et al. Randomized Comparison of Low-Molecular-Weight Heparin versus Oral Anticoagulant Therapy for the Prevention of Recurrent Venous Thromboembolism in Patients with Cancer (CLOT) Investigators. Low-molecular-weight heparin versus a coumarin for the prevention of recurrent venous thromboembolism in patients with cancer. N Engl J Med. 2003;349:146-53. [PMID: 12853587] 8. Lee AY, Rickles FR, Julian JA, Gent M, Baker RI, Bowden C, et al. Randomized comparison of low molecular weight heparin and coumarin derivatives on the survival of patients with cancer and venous thromboembolism. J Clin Oncol. 2005;23:2123-9. [PMID: 15699480] 9. Lazo-Langner A, Goss GD, Spaans JN, Rodger MA. The effect of lowmolecular-weight heparin on cancer survival. A systematic review and meta-analysis of randomized trials. J Thromb Haemost. 2007;5:729-37. [PMID: 17408406] 10. Nordstrom M, Lindblad B, Anderson H, Bergqvist D, Kjellstrom T. Deep venous thrombosis and occult malignancy: an epidemiological study. BMJ. 1994; 308:891-4. [PMID: 8173368] 11. Cornuz J, Pearson SD, Creager MA, Cook EF, Goldman L. Importance of ndings on the initial evaluation for cancer in patients with symptomatic idiopathic deep venous thrombosis. Ann Intern Med. 1996;125:785-93. [PMID: 8928984] 12. Monreal M, Lensing AW, Prins MH, Bonet M, Fernandez-Llamazares J, Muchart J, et al. Screening for occult cancer in patients with acute deep vein thrombosis or pulmonary embolism. J Thromb Haemost. 2004;2:876-81. [PMID: 15140120] 13. Piccioli A, Lensing AW, Prins MH, Falanga A, Scannapieco GL, Ieran M, et al. SOMIT Investigators Group. Extensive screening for occult malignant disease in idiopathic venous thromboembolism: a prospective randomized clinical trial. J Thromb Haemost. 2004;2:884-9. [PMID: 15140122] 14. Wells GA, Shea B, OConnell D, Peterson J, Welch V, Losos M, et al. The NewcastleOttawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Accessed at www.ohri.ca/programs/clinical_epidemiology /oxford.htm on 11 July 2008. 15. Khan KS, ter Riet G, Popay J, Nixon J, Kleijnen J. Phase 5: Study quality assessment. In: CRD Report 4 Undertaking Systematic Reviews of Research on Effectiveness: CRDs Guidance for Carrying Out or Commissioning Reviews. 2nd ed. Heslington, York: University of York, NHS Centre for Reviews and Dissemination; 2001. 16. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA. 1995;273:408-12. [PMID: 7823387] 17. Bura A, Cailleux N, Bienvenu B, Leger P, Bissery A, Boccalon H, et al. Incidence and prognosis of cancer associated with bilateral venous thrombosis: a prospective study of 103 patients. J Thromb Haemost. 2004;2:441-4. [PMID: 15009461] 18. Monreal M, Fernandez-Llamazares J, Perandreu J, Urrutia A, Sahuquillo JC, Contel E. Occult cancer in patients with venous thromboembolism: which patients, which cancers. Thromb Haemost. 1997;78:1316-8. [PMID: 9408011] 19. Hettiarachchi RJ, Lok J, Prins MH, Buller HR, Prandoni P. Undiagnosed malignancy in patients with deep vein thrombosis: incidence, risk indicators, and diagnosis. Cancer. 1998;83:180-5. [PMID: 9655309] 20. Aderka D, Brown A, Zelikovski A, Pinkhas J. Idiopathic deep vein thromwww.annals.org

Epidemiology Program, Ottawa Health Research Institute, 501 Smyth Road, Room W6116, Eye Institute, Ottawa, Ontario K1H 8L6, Canada; e-mail, mrodger@ohri.ca. Current author addresses are available at www.annals.org.
332 2 September 2008 Annals of Internal Medicine Volume 149 Number 5

Cancer Screening in Patients with Venous Thromboembolism

bosis in an apparently healthy patient as a premonitory sign of occult cancer. Cancer. 1986;57:1846-9. [PMID: 3955523] 21. Prandoni P, Lensing AW, Buller HR, Cogo A, Prins MH, Cattelan AM, et al. Deep-vein thrombosis and the incidence of subsequent symptomatic cancer. N Engl J Med. 1992;327:1128-33. [PMID: 1528208] 22. Rajan R, Levine M, Gent M, Hirsh J, Geerts W, Skingley P, et al. The occurrence of subsequent malignancy in patients presenting with deep vein thrombosis: results from a historical cohort study. Thromb Haemost. 1998;79: 19-22. [PMID: 9459315] 23. Ahmed Z, Mohyuddin Z. Deep vein thrombosis as a predictor of cancer. Angiology. 1996;47:261-5. [PMID: 8638869] 24. Subira M, Mateo J, Souto JC, Altes A, Fontcuberta J. Lack of association ` between venous thrombosis and subsequent malignancy in a retrospective cohort study in young patients. Am J Hematol. 1999;60:181-4. [PMID: 10072107] 25. Gore JM, Appelbaum JS, Greene HL, Dexter L, Dalen JE. Occult cancer in patients with acute pulmonary embolism. Ann Intern Med. 1982;96:556-60. [PMID: 7073147] 26. Girolami A, Prandoni P, Zanon E, Bagatella P, Girolami B. Venous thromboses of upper limbs are more frequently associated with occult cancer as compared with those of lower limbs. Blood Coagul Fibrinolysis. 1999;10:455-7. [PMID: 10636455] 27. Monreal M, Salvador R, Soriano V, Sabria M. Cancer and deep venous thrombosis [Letter]. Arch Intern Med. 1988;148:485. [PMID: 3341848] 28. OConnor NT, Cederholm-Williams SA, Fletcher EW, Allington M, Sharp AA. Signicance of idiopathic deep venous thrombosis. Postgrad Med J. 1984; 60:275-7. [PMID: 6539470] 29. Goldberg RJ, Seneff M, Gore JM, Anderson FA Jr, Greene HL, Wheeler HB, et al. Occult malignant neoplasm in patients with deep venous thrombosis. Arch Intern Med. 1987;147:251-3. [PMID: 3813741] 30. Grifn MR, Stanson AW, Brown ML, Hauser MF, OFallon WM, Anderson HM, et al. Deep venous thrombosis and pulmonary embolism. Risk of subsequent malignant neoplasms. Arch Intern Med. 1987;147:1907-11. [PMID: 3675090] 31. Rance A, Emmerich J, Guedj C, Fiessinger JN. Occult cancer in patients with bilateral deep-vein thrombosis [Letter]. Lancet. 1997;350:1448-9. [PMID: 9371175] 32. Ranft J, Heidrich H. Frequency of malignant diseases in deep venous thrombosis of the lower extremities. Int Angiol. 1991;10:66-8. [PMID: 1861088] 33. Minar E, Ehringer H, Marosi L, Hirschi M, Konecny U, Pollak C, et al. Acute venous thrombosis: localization, extent and aetiology with special consideration of paraneoplasia. Dtsch Med Wochenschr. 1982;107:1303-9. [PMID: 7117140] 34. Saba HI, Khalil FK, Morelli GA, Foulis PR. Thromboembolism preceding cancer: a correlation study. Am J Hematol. 2003;72:109-14. [PMID: 12555214] 35. Monreal M, Lafoz E, Casals A, Inaraja L, Montserrat E, Callejas JM, et al. Occult cancer in patients with deep venous thrombosis. A systematic approach. Cancer. 1991;67:541-5. [PMID: 1985747] 36. Sannella NA, OConnor DJ Jr. Idiopathic deep venous thrombosis: the value of routine abdominal and pelvic computed tomographic scanning. Ann Vasc Surg. 1991;5:218-22. [PMID: 2064913] 37. Barrellier MT, Carpentier P, Henriet JP, Lequerrec A, Derlon J, Khayat MC, et al. [Prospective etiologic study of 128 patients with ambulatory dep vein thrombosis]. J Mal Vasc. 1992;17:277-83. [PMID: 1494055]

Review

38. Monreal M, Casals A, Boix J, Olazabal A, Montserrat E, Mundo MR. Occult cancer in patients with acute pulmonary embolism. A prospective study. Chest. 1993;103:816-9. [PMID: 8449074] 39. Pistorius MA, Le Menner G, Planchon B. [Deep venous thrombosis and cancer. A one year evaluation of etiologic screening of patients hospitalized in Internal Medicine]. J Mal Vasc. 1994;19:273-7. [PMID: 7852870] 40. Bastounis EA, Karayiannakis AJ, Makri GG, Alexiou D, Papalambros EL. The incidence of occult cancer in patients with deep venous thrombosis: a prospective study. J Intern Med. 1996;239:153-6. [PMID: 8568483] 41. Cailleux N, Marie I, Primard E, Lecomte F, Henry J, Louvel JP, et al. [Thrombophlebitis and cancer: evaluation of the diagnostic value of abdominal ultrasonography in the acute phase of a deep venous thrombosis. Report of 148 consecutive examinations]. J Mal Vasc. 1997;22:322-5. [PMID: 9479602] 42. Fahrig C, Heidrich H, Penninger C. The incidence of occult malignant diseases in patients with deep venous thrombosis of the pelvis and lower limb. Int J Angiol. 1998;7:249-51. [PMID: 9585461] 43. Netzer P, Binek J, Hammer B, Schmassmann A. Utility of abdominal sonography in patients with idiopathic deep vein thrombosis. J Clin Ultrasound. 1999;27:177-81. [PMID: 10323187] 44. Ronsdorf A, Perruchoud AP, Schoenenberger RA. Search for occult malignancy in patients with deep venous thrombosis. Results of a retrospective cohort study. Swiss Med Wkly. 2003;133:567-74. [PMID: 14691729] 45. van Doormaal FF, Otten H, Buller HR. Screening for occult malignancy in patients with idiopathic venous thromboembolism. The Trousseau Investigators. J Thromb Haemost. 2007;5(Suppl 2):Abstract P-T-501. 46. Engudanos MJ, Todol JA, Saro E, Salvador G, Villar J, Gomez-Biedma S, et al. [Usefulness of the tumor markers in the diagnosis of idiopathic deep venous thrombosis associated cancer]. An Med Interna. 2002;19:561-6. [PMID: 12522891] 47. Halici U, Cikirikcioglu M, Canbaz S, Ketenciler S, Duran E. The role of abdominopelvic ultrasonography in detecting the occult malignances in patients with lower limb venous thrombosis. Trakya Univ Fak Derg. 2006;23:9-13. 48. Biland L, Zemp E, Widmer LK. [Epidemiology of venous thromboembolism]. Internist (Berl). 1987;28:285-90. [PMID: 3301717] 49. Prandoni P, Lensing AW, Piccioli A, Bernardi E, Simioni P, Girolami B, et al. Recurrent venous thromboembolism and bleeding complications during anticoagulant treatment in patients with cancer and venous thrombosis. Blood. 2002; 100:3484-8. [PMID: 12393647] 50. Snowdon C, Elbourne D, Garcia J. Zelen randomization: attitudes of parents participating in a neonatal clinical trial. Control Clin Trials. 1999;20:14971. [PMID: 10227415] 51. Buller HR, van Doormaal FF, van Sluis GL, Kamphuisen PW. Cancer and thrombosis: from molecular mechanisms to clinical presentations. J Thromb Haemost. 2007;5 Suppl 1:246-54. [PMID: 17635733] 52. Piccioli A, Prandoni P. Screening for occult cancer in patients with idiopathic venous thromboembolism: yes. J Thromb Haemost. 2003;1:2271-2. [PMID: 14629455] 53. Lee AY. Screening for occult cancer in patients with idiopathic venous thromboembolism: no. J Thromb Haemost. 2003;1:2273-4. [PMID: 14629456] 54. Lee AY. Thrombosis and cancer: the role of screening for occult cancer and recognizing the underlying biological mechanisms. Hematology Am Soc Hematol Educ Program. 2006:438-43. [PMID: 17124096]

www.annals.org

2 September 2008 Annals of Internal Medicine Volume 149 Number 5 333

Current Author Addresses: Drs. Carrier, Fergusson, and Ramsay: Eye

Institute, 501 Smyth Road, Room W6109, Ottawa, Ontario K1H 8L6, Canada. Dr. Le Gal: Centre Hospitalier Universitaire de la Cavale Blanche, EA 3878, 29609 Brest, France. Dr. Wells: Ottawa Hospital, Civic Campus, Suite F6-49, 1053 Carling Avenue, Ottawa, Ontario K1Y 4E9, Canada.

Appendix Table 1. MEDLINE Search Strategy

1. exp Thrombosis/ 2. exp Venous Thrombosis/ 3. Thromb$.ti,ab. 4. phlebothrombosis.mp. 5. Deep vein thrombosis.mp. 6. pulmonary emboli.mp. 7. venous thromboembolic event$.mp. 8. or/1-7 9. exp Neoplasms/ 10. cancer.mp. 11. neopl$.ti,ab. 12. tumor$.ti,ab. 13. cancer$.ti,ab. 14. exp Medical Oncology/ 15. onco$.ti,ab. 16. exp Hematologic Neoplasms/ 17. or/9-16 18. exp Early Diagnosis/ 19. exp Mass Screening/ 20. screening.mp. 21. or/18-20 22. 8 and 17 and 21

Appendix Table 2. Quality of Included Studies Assessed by Using the NewcastleOttawa Quality Scale for Cohort Studies*
Author, Year (Reference) Representativeness of Exposed Cohort Selection Representativeness of Nonexposed Cohort NA Ascertainment of Exposure Outcome Not Present at Beginning of Study NA Comparability of Cohorts Assessment of Outcome Outcome Was Follow-Up Long Enough? Adequacy of Follow-Up

Sannella and OConnor, 1991 (36) Fahrig et al., 1998 (42) Netzer et al., 1999 (43) Ronsdorf et al., 2003 (44) Monreal et al., 1991 (35) Barrelier et al., 1992 (37) Monreal et al., 1993 (38) Pistorius et al., 1994 (39) Bastounis et al., 1996 (40) Cailleux et al., 1997 (41) Monreal et al., 1997 (18) Engudanos et al., 2002 (46) Monreal et al., 2004 (12) van Doormaal et al., 2007 (45) Piccioli et al., 2004 (13)

NA NA NA NA NA NA NA NA NA NA NA NA

NA NA NA NA NA NA NA NA NA NA NA NA

NA not applicable. *A study can be awarded a maximum of 1 star for each numbered item in the selection and outcome categories. A maximum of 2 stars can be given for comparability. Randomized, controlled trial.
W-68 2 September 2008 Annals of Internal Medicine Volume 149 Number 5 www.annals.org

Appendix Table 3. Frequency of Various Types of Cancer after VTE Diagnosis*

Cancer Type (Total Frequency) Monreal et al., 1991 (35) Gastric (n 17) Pancreatic (n 25) Colorectal (n 49) Esophageal (n 1) Liver/biliary tree (n 7) Lung (n 32) Bladder (n 14) Kidney (n 12) Ovarian (n 13) Uterine/cervical (n 10) Prostate (n 36) Testicular (n 1) Breast (n 14) CLL (n 3) Multiple myeloma (n 1) Lymphoma (n 10) MPD (n 6) Acute leukemia (n 6) Sarcoma (n 2) ENT (n 7) Hemangioendothelioma (n Brain (n 1) Adrenal (n 1) Skin (n 3) Unknown (n 14) Not defined (n 60) 2 1 2 0 0 1 2 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 1 1 0 Sannella and OConnor, 1991 (36) 0 2 1 0 1 1 0 1 3 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 Barrelier et al., 1992 (37) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 1 0 Monreal et al., 1993 (38) 0 0 3 1 1 0 0 0 1 0 2 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 Pistorius et al., 1994 (39) 0 0 2 0 0 0 0 0 0 1 2 0 1 0 0 0 1 0 0 0 0 0 0 1 1 0 Bastounis et al., 1996 (40) 2 5 10 0 0 2 1 1 1 1 3 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 Cailleux et al., 1997 (41) 0 1 1 0 1 0 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 Cases of Cancer, n Monreal et al., 1997 (18) 3 2 6 0 1 1 2 1 2 1 9 0 1 0 0 2 0 0 0 2 0 0 0 0 1 0 Fahrig et al., 1998 (42) 1 2 3 0 0 4 1 3 1 1 2 0 1 1 0 2 0 1 0 0 1 0 0 0 0 0 Netzer et al., 1999 (43) 0 2 3 0 0 0 2 0 0 0 1 0 2 0 0 1 0 0 0 0 0 0 0 0 3 0 Engudanos et al., 2002 (46) 0 1 1 0 0 4 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 Ronsdorf et al., 2003 (44) 1 3 7 0 0 8 0 2 1 0 5 0 3 0 0 3 2 1 0 3 0 0 0 1 2 0 Monreal et al., 2004 (12) 6 3 7 0 2 7 2 3 3 3 10 0 2 0 0 1 1 3 1 2 0 0 0 0 5 0 Piccioli et al., 2004 (13) 2 3 3 0 1 4 3 1 1 1 2 0 2 0 0 0 0 0 0 0 0 0 1 0 0 32 van Doormaal et al., 2007 (45) NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 28

1)

CLL chronic lymphocytic leukemia; ENT * Based on studies reporting cancer types.

ear, nose, and throat; MPD

myeloproliferative disorder; NA

not applicable; VTE

venous thromboembolism.

www.annals.org

2 September 2008 Annals of Internal Medicine Volume 149 Number 5 W-69