Post on 22-Sep-2020
transcript
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
Cancer and thrombosisImprovements in strategies for prediction, diagnosis, and treatmentvan Es, N.
Link to publication
LicenseOther
Citation for published version (APA):van Es, N. (2017). Cancer and thrombosis: Improvements in strategies for prediction, diagnosis, and treatment.
General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.
Download date: 17 Jan 2021
CHAPTER 10
SCREENING FOR OCCULT CANCER IN PATIENTS WITH UNPROVOKED VENOUS
THROMBOEMBLISM: A SYSTEMATIC REVIEW AND INDIVIDUAL PATIENT DATA
META-ANALYSIS
Nick van Es, Grégoire Le Gal, Hans-Martin Otten, Philippe Robin, Andrea Piccioli, Ramón Lecumberri, Luis Jara-Palomares, Piotr Religa, Virginie Rieu, Matthew T. Rondina,
Mariëlle M. Beckers, Paolo Prandoni, Pierre-Yves Salaun, Marcello Di Nisio, Patrick M. Bossuyt, Harry R. Büller, and Marc Carrier
Annals of Internal Medicine, 2017
Chapter 10
184
ABSTRACT
BACKGROUND Screening for cancer in patients with unprovoked venous thromboembolism (VTE) is often considered, but clinicians need precise data on cancer prevalence, risk factors, and effect of different types of screening strategies.
PURPOSE To estimate the prevalence of occult cancer in patients with unprovoked VTE including in subgroups of different age or that have undergone different types of screening.
DATA SOURCES MEDLINE, Embase, and CENTRAL up to January 19th, 2016.
STUDY SELECTION Prospective studies evaluating cancer screening strategies in adults with unprovoked VTE, with enrolment starting after January 1st, 2000, and at least 12 months follow‐up.
DATA EXTRACTION AND SYNTHESIS Two investigators independently reviewed abstracts and full‐text articles and independently assessed risk of bias.
DATA SYNTHESIS Ten eligible studies were identified. Individual data were obtained for all 2,316 patients. The mean age was 60 years; 58% received extensive screening. The 12‐month period prevalence of cancer following the VTE diagnosis was 5.2% (95% CI, 4.1 to 6.5). The point prevalence of cancer was statistically significantly higher in patients receiving extensive screening than those receiving a more limited screening at initial screening (odds ratio 2.0 [OR]; 95% CI, 1.2 to 3.4), but not at 12 months (OR 1.4; 95% CI, 0.89 to 2.1). Cancer prevalence increased linearly with age, and was seven‐fold higher in patients of 50 years or older than in younger patients (OR 7.1; 95% CI, 3.1 to 16).
LIMITATIONS Variation in patient characteristics and extensive screening strategies; long‐term mortality data not available.
CONCLUSIONS Occult cancer is detected in one of every twenty patients within a year of their unprovoked VTE diagnosis. Older age was associated with a higher cancer prevalence. Although an extensive screening strategy may initially detect more cancers than limited screening, it remains unclear whether this translates into improved patient important outcomes.
Screening for occult cancer in unprovoked venous thromboembolism
185
10
ABSTRACT
BACKGROUND Screening for cancer in patients with unprovoked venous thromboembolism (VTE) is often considered, but clinicians need precise data on cancer prevalence, risk factors, and effect of different types of screening strategies.
PURPOSE To estimate the prevalence of occult cancer in patients with unprovoked VTE including in subgroups of different age or that have undergone different types of screening.
DATA SOURCES MEDLINE, Embase, and CENTRAL up to January 19th, 2016.
STUDY SELECTION Prospective studies evaluating cancer screening strategies in adults with unprovoked VTE, with enrolment starting after January 1st, 2000, and at least 12 months follow‐up.
DATA EXTRACTION AND SYNTHESIS Two investigators independently reviewed abstracts and full‐text articles and independently assessed risk of bias.
DATA SYNTHESIS Ten eligible studies were identified. Individual data were obtained for all 2,316 patients. The mean age was 60 years; 58% received extensive screening. The 12‐month period prevalence of cancer following the VTE diagnosis was 5.2% (95% CI, 4.1 to 6.5). The point prevalence of cancer was statistically significantly higher in patients receiving extensive screening than those receiving a more limited screening at initial screening (odds ratio 2.0 [OR]; 95% CI, 1.2 to 3.4), but not at 12 months (OR 1.4; 95% CI, 0.89 to 2.1). Cancer prevalence increased linearly with age, and was seven‐fold higher in patients of 50 years or older than in younger patients (OR 7.1; 95% CI, 3.1 to 16).
LIMITATIONS Variation in patient characteristics and extensive screening strategies; long‐term mortality data not available.
CONCLUSIONS Occult cancer is detected in one of every twenty patients within a year of their unprovoked VTE diagnosis. Older age was associated with a higher cancer prevalence. Although an extensive screening strategy may initially detect more cancers than limited screening, it remains unclear whether this translates into improved patient important outcomes.
INTRODUCTION Unprovoked venous thromboembolism (VTE) may be the first sign of occult cancer. Screening is often considered in these patients with the aim to detect underlying cancers at an early, curable stage and to reduce cancer-related morbidity and mortality. The extent to which patients with unprovoked VTE should be screened for occult cancer is controversial. Although an early study suggested that an extensive screening strategy may detect more cancers than a more limited one1 recent studies evaluating extensive screening strategies using computed tomography (CT) of the abdomen2–4 or whole-body positron emission tomography (PET)/CT5 could not confirm this finding. Extensive screening tests may yield false-positive findings, requiring additional, sometimes invasive testing, which increases healthcare costs, exposes patients to potential procedure-related complications, and can lead to patient anxiety. Given the lack of a clear benefit and the potential harms of extensive screening, a more limited occult cancer screening strategy is currently suggested in patients with unprovoked VTE,6 comprising of medical history, physical examination, basic blood work, a chest radiograph, and age- and gender-specific testing. To guide decisions about occult cancer screening and to counsel patients, clinicians need precise estimates of the period prevalence of occult cancer at the time of VTE diagnosis and during follow-up. Clinicians also need to be informed about the types and stage of cancer that can potentially be detected by screening and about the tests that are useful for occult cancer detection. To help clinicians tailor screening decisions, we performed a systematic review and an individual patient data meta-analysis, combining patient-level data from ten recently published, prospective studies of occult cancer screening in patients with unprovoked VTE.
METHODS Methods were pre-specified in a protocol that has previously been published.7 The guidance of the PRISMA-IPD (Preferred Reporting Items for Systematic reviews and Meta-Analyses of Individual Participant Data) Statement was followed.8
Data sources and searches Our previously published systematic review9 was updated by searching MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials databases from November 1st, 2007
Chapter 10
186
up to January 19th, 2016, combining terms for venous thromboembolism, cancer, and screening (Table 1). No language restrictions were applied. In addition, conference proceedings of the International Society on Thrombosis and Haemostasis and the American Society of Hematology were searched from 2007 to 2015. Reference lists of eligible articles were hand-searched. Two authors (NvE and MC) independently screened the titles and abstracts of articles and assessed the full-texts for eligibility. Any differences of opinion regarding study eligibility were resolved by discussion. The systematic review was registered with the International Prospective Registry of Systematic Reviews (PROSPERO; CRD42016033371).
Table 1. MEDLINE search strategy
Search string 1 “Thrombosis”[MeSH] OR “pulmonary embolism”[MeSH] OR thrombos*[tiab] OR
thrombot*[tiab] OR thromboemboli*[tiab] OR phlebothrombosis[All Fields] OR “deep vein thrombosis”[All Fields] OR pulmonary emboli*[All Fields] OR venous thromboembolic event*[All Fields]
2 Neoplasms[MeSH] OR cancer[All Fields] OR neopl*[tiab] OR tumor*[tiab] OR tumour*[tiab] OR cancer*[tiab] OR “Medical Oncology”[MeSH] OR oncol*[tiab] OR “Hematologic Neoplasms”[MeSH]
3 “Early Diagnosis”[MeSH] OR “Mass Screening”[MeSH] OR screen*[tiab] OR occult*[tiab] 4 "2007/11/01"[PDAT] : "3000/12/31"[PDAT] 5 1 AND 2 AND 3 AND 4
Study selection To be eligible, studies had to have prospectively included consecutive adult patients with unprovoked, objectively confirmed deep vein thrombosis or pulmonary embolism and followed them for a minimum of 12 months for potential cancer. We accepted the individual study definitions of unprovoked VTE. Studies were required to follow a defined strategy for occult cancer screening, including at least a medical history, physical examination, basic blood tests, and chest X-ray. Studies that had started patient enrolment before January 1, 2000 were excluded. This was done because differences in cancer screening practices (e.g. age- and gender-specific testing) and in the performance of diagnostic tests were felt to be significant, and more recent studies were considered more relevant and informative. All studies that had enrolled patients prior to any screening procedures were included in the primary analyses. Studies enrolling patients only after a negative initial screening strategy were used for additional analyses.
Screening for occult cancer in unprovoked venous thromboembolism
187
10
up to January 19th, 2016, combining terms for venous thromboembolism, cancer, and screening (Table 1). No language restrictions were applied. In addition, conference proceedings of the International Society on Thrombosis and Haemostasis and the American Society of Hematology were searched from 2007 to 2015. Reference lists of eligible articles were hand-searched. Two authors (NvE and MC) independently screened the titles and abstracts of articles and assessed the full-texts for eligibility. Any differences of opinion regarding study eligibility were resolved by discussion. The systematic review was registered with the International Prospective Registry of Systematic Reviews (PROSPERO; CRD42016033371).
Table 1. MEDLINE search strategy
Search string 1 “Thrombosis”[MeSH] OR “pulmonary embolism”[MeSH] OR thrombos*[tiab] OR
thrombot*[tiab] OR thromboemboli*[tiab] OR phlebothrombosis[All Fields] OR “deep vein thrombosis”[All Fields] OR pulmonary emboli*[All Fields] OR venous thromboembolic event*[All Fields]
2 Neoplasms[MeSH] OR cancer[All Fields] OR neopl*[tiab] OR tumor*[tiab] OR tumour*[tiab] OR cancer*[tiab] OR “Medical Oncology”[MeSH] OR oncol*[tiab] OR “Hematologic Neoplasms”[MeSH]
3 “Early Diagnosis”[MeSH] OR “Mass Screening”[MeSH] OR screen*[tiab] OR occult*[tiab] 4 "2007/11/01"[PDAT] : "3000/12/31"[PDAT] 5 1 AND 2 AND 3 AND 4
Study selection To be eligible, studies had to have prospectively included consecutive adult patients with unprovoked, objectively confirmed deep vein thrombosis or pulmonary embolism and followed them for a minimum of 12 months for potential cancer. We accepted the individual study definitions of unprovoked VTE. Studies were required to follow a defined strategy for occult cancer screening, including at least a medical history, physical examination, basic blood tests, and chest X-ray. Studies that had started patient enrolment before January 1, 2000 were excluded. This was done because differences in cancer screening practices (e.g. age- and gender-specific testing) and in the performance of diagnostic tests were felt to be significant, and more recent studies were considered more relevant and informative. All studies that had enrolled patients prior to any screening procedures were included in the primary analyses. Studies enrolling patients only after a negative initial screening strategy were used for additional analyses.
Data extraction and quality assessment Corresponding authors of eligible studies were invited to participate in this collaborative project. All contacted authors agreed and provided individual patient data. Study-level information was sought for the different study aims, definition of unprovoked VTE, screening strategy applied, follow-up duration, and assessment of outcomes. Patient-level data about baseline characteristics, risk factors, index VTE, cancer screening tests, and outcomes, including cancer, recurrent VTE, death, and loss to follow-up, were obtained. Patients enrolled more than 90 days after the VTE diagnosis were excluded from the dataset, as well as patients in whom the index VTE was not objectively confirmed. To ensure data consistency, baseline tables and primary analyses reported in the original articles were reconstructed. Discrepancies with the published tables were resolved by contacting the principal investigators. All studies had been approved by the institutional review boards of participating centers. Two reviewers (NvE and NK) independently assessed the potential risks of bias for each study using the Newcastle-Ottawa Scale10 as well as Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool, which was adapted to the present research question.11 Disagreements were resolved by consensus.
Data synthesis and analysis The primary outcome measure was the point prevalence of previously undiagnosed cancer in patients with unprovoked VTE at 12 months, defined as the proportion of patients in whom solid or hematological cancer (excluding non-melanoma skin cancer) was objectively confirmed by histology or cytology, or unequivocally diagnosed by imaging or tumor markers. Cancer detection refers to cancers detected by screening tests, which were subsequently confirmed by additional testing. Cancer diagnosis refers to all confirmed cancers, either at screening or during follow-up. The period prevalence of cancer was also analysed at different time points including initial screening, between screening and 12 months, and between 12 and 24 months of follow-up. In subgroup analyses, the effects of type of screening (limited vs. extensive) and of patients’ age (by cohorts of 10 years) on the probability of cancer diagnosis were assessed. Finally, the probability of a cancer diagnosis was estimated in males, females, current or former smokers, women on estrogen-containing oral contraceptive or hormone replacement therapy, patients presenting with PE, and those with previous VTE.
Chapter 10
188
Secondary outcomes were early-stage solid cancers (stage 0, 1 or 2 according to the American Joint Committee on Cancer staging system), stage 3 and 4 solid cancers, and haematological malignancies. The proportion of positive findings at limited screening requiring additional testing and the proportion of patients subsequently diagnosed with cancer were evaluated. Limited screening was defined as the combination of medical history, physical examination, basic blood tests (complete blood count, creatinine, and liver function tests), a chest X-ray, and/or age- and gender-specific testing, such as mammography or prostate specific antigen. Limited screening was considered to be positive if results led to additional investigations for possible cancer detection. Extensive screening strategies were heterogeneous across the studies, but often included imaging with a CT of the abdomen, ultrasound of the abdomen, or whole-body PET/CT (Table 2). Patient-level information was obtained from source documentation for each cancer diagnosed at initial screening. Using detailed narratives based on source documents, three authors (NvE, GLG, and MC) independently adjudicated which of the screening tests initially had raised the suspicion of cancer and eventually led to cancer detection.
Screening for occult cancer in unprovoked venous thromboembolism
189
10
Secondary outcomes were early-stage solid cancers (stage 0, 1 or 2 according to the American Joint Committee on Cancer staging system), stage 3 and 4 solid cancers, and haematological malignancies. The proportion of positive findings at limited screening requiring additional testing and the proportion of patients subsequently diagnosed with cancer were evaluated. Limited screening was defined as the combination of medical history, physical examination, basic blood tests (complete blood count, creatinine, and liver function tests), a chest X-ray, and/or age- and gender-specific testing, such as mammography or prostate specific antigen. Limited screening was considered to be positive if results led to additional investigations for possible cancer detection. Extensive screening strategies were heterogeneous across the studies, but often included imaging with a CT of the abdomen, ultrasound of the abdomen, or whole-body PET/CT (Table 2). Patient-level information was obtained from source documentation for each cancer diagnosed at initial screening. Using detailed narratives based on source documents, three authors (NvE, GLG, and MC) independently adjudicated which of the screening tests initially had raised the suspicion of cancer and eventually led to cancer detection.
Table 2. Study
characteris
tics
First a
utho
r Aim
Desig
n Ce
nters
Stud
y pe
riod
Follo
w‐up
N Ca
ncer
Enrolm
ent p
rior to screen
ing proced
ures
Carrie
r et a
l. (201
0)
To evaluate an
exten
sive CT
‐ba
sed screen
ing strategy
Prospe
ctive
coho
rt
Sing
le
2006
‐200
7 24
mon
ths
53
2
Carrie
r et a
l. (201
5)
To co
mpa
re abd
ominop
elvic
CT with
limite
d screen
ing
RCT
Multi
2008
‐201
4 12
mon
ths
854
33
Jara‐Pa
lomares et a
l.
To evaluate of an extensive
screen
ing strategy
Prospe
ctive
coho
rt
Sing
le
2003
‐200
4 24
mon
ths
49
3
Rieu et al.
To evaluate an
exten
sive
screen
ing strategy
Prospe
ctive
coho
rt
Sing
le
2004
‐200
5 12
mon
ths
32
4
Rob
in et a
l. To
compa
re PET
/CT with
a
limite
d screen
ing strategy
RCT
Multi
2009
‐201
2 24
mon
ths
394
25
Ron
dina
et a
l. To
evaluate PE
T/CT
‐based
screen
ing strategy
Prospe
ctive
coho
rt
Sing
le
2008
‐201
0 24
mon
ths
40
1
Van
Doo
rmaal et a
l. To
compa
re CT of th
e chest
and ab
domen
with
a limite
d screen
ing strategy
CCT
Multi
2002
‐200
7 Med
ian 2.5 y
630
50
Enrolm
ent a
fter n
egative lim
ited screen
ing
Alfo
nso et al.
To evaluate PE
T/CT
in
patie
nts w
ith a negative
limite
d screen
ing
Prospe
ctive
coho
rt
Sing
le
2007
‐201
0 24
mon
ths
99
7
Beckers et a
l.
To evaluate PE
T/CT
in
patie
nts w
ith a negative
limite
d screen
ing
Prospe
ctive
coho
rt
Sing
le
2003
‐200
6 24
mon
ths
25
0
Prand
oni et a
l.
To co
mpa
re abd
ominal CT
with
a co
mmon
clinica
l practic
e ap
proa
ch
RCT
Multi
2006
‐200
8 24
mon
ths
205
18
Abbreviatio
ns: C
CT, con
curren
tly co
ntrolled trial; CT
, com
puted tomog
raph
y; PET, p
ositron
emiss
ion tomog
raph
y; RCT
, ran
domize
d controlled trial.
Chapter 10
190
Statistical Analysis
Summary probability estimates were obtained in one-stage meta-analysis, using a generalized linear mixed-effects model, in which a study-specific random effect was included to account for the clustering of observations within studies. Subgroup differences were analysed with an indicator variable, as a fixed effect. When analyzing the effect of type of screening, we added a random effect because of the differences in extensive screening strategies across studies and adjusted the analysis for age, sex, smoking, and index VTE, since one of the comparative studies was non-randomized. Marginal probability estimates were calculated by integrating the inverse logit of the fixed effect of the intercept over the random-effects distribution, using Gauss-Hermite quadrature approximation, with 10 quadrature points to arrive at the final point estimates and 95% confidence intervals (CI). To illustrate heterogeneity across the studies, 95% prediction intervals (PIs) were calculated around the point estimates, based on the standard error of the fixed effect and the variance of the random effect. Forest plots were also generated to visualize potential heterogeneity. We additionally performed a predefined sensitivity analysis restricted to patients enrolled within 30 days following the index VTE. All analyses were undertaken on an intention-to-screen basis and performed in R, version 3.3.2 (R Foundation for Statistical Computing; www.R-project.org) using the lme4 version 1.1-12 package for the mixed-effects models. A significance level of 0.05 was used; all tests were two-tailed.
RESULTS Included studies A total of 1,216 articles were identified, of which 23 full-texts were assessed for eligibility (Figure 1). Ten prospective studies met the inclusion criteria and individual patient data for all 2,371 patients were obtained.2–5,12–17 Seven of these studies (N=2,052) had enrolled patients prior to any occult cancer screening procedures.5,14–18 Study characteristics are depicted in Table 2. The definition of unprovoked VTE was consistent across studies and mostly encompassed symptomatic, objectively confirmed acute pulmonary embolism and/or deep vein thrombosis in the absence of known cancer, recent surgery, trauma of the leg, immobilization, previous unprovoked VTE, known thrombophilia, pregnancy or puerperium. Study group size ranged from 25 to 854 patients. Patients were enrolled between
Screening for occult cancer in unprovoked venous thromboembolism
191
10
Statistical Analysis
Summary probability estimates were obtained in one-stage meta-analysis, using a generalized linear mixed-effects model, in which a study-specific random effect was included to account for the clustering of observations within studies. Subgroup differences were analysed with an indicator variable, as a fixed effect. When analyzing the effect of type of screening, we added a random effect because of the differences in extensive screening strategies across studies and adjusted the analysis for age, sex, smoking, and index VTE, since one of the comparative studies was non-randomized. Marginal probability estimates were calculated by integrating the inverse logit of the fixed effect of the intercept over the random-effects distribution, using Gauss-Hermite quadrature approximation, with 10 quadrature points to arrive at the final point estimates and 95% confidence intervals (CI). To illustrate heterogeneity across the studies, 95% prediction intervals (PIs) were calculated around the point estimates, based on the standard error of the fixed effect and the variance of the random effect. Forest plots were also generated to visualize potential heterogeneity. We additionally performed a predefined sensitivity analysis restricted to patients enrolled within 30 days following the index VTE. All analyses were undertaken on an intention-to-screen basis and performed in R, version 3.3.2 (R Foundation for Statistical Computing; www.R-project.org) using the lme4 version 1.1-12 package for the mixed-effects models. A significance level of 0.05 was used; all tests were two-tailed.
RESULTS Included studies A total of 1,216 articles were identified, of which 23 full-texts were assessed for eligibility (Figure 1). Ten prospective studies met the inclusion criteria and individual patient data for all 2,371 patients were obtained.2–5,12–17 Seven of these studies (N=2,052) had enrolled patients prior to any occult cancer screening procedures.5,14–18 Study characteristics are depicted in Table 2. The definition of unprovoked VTE was consistent across studies and mostly encompassed symptomatic, objectively confirmed acute pulmonary embolism and/or deep vein thrombosis in the absence of known cancer, recent surgery, trauma of the leg, immobilization, previous unprovoked VTE, known thrombophilia, pregnancy or puerperium. Study group size ranged from 25 to 854 patients. Patients were enrolled between
October 2002 and April 2014, in Europe.2,5,12–14,16,19 and in North America.3,15,17 Median follow-up duration ranged from one year to 2.5 years (overall median 500 days). Patients were contacted at fixed time intervals to elicit information about a new cancer diagnosis in eight studies.2,4,5,13–15,17,18 Testing for cancer during follow-up was left to the discretion of the physician in all studies.
Figure 1. PRISMA flow chart of systematic review
Of the seven studies that enrolled patients prior to any screening procedures, three compared a limited screening strategy with an extensive screening strategy (N=1,878),2,3,5 while four studies only evaluated an extensive screening strategy (N=174) (Table 2).14–17 The three other studies (N=319) included patients after a negative limited screening; two evaluated PET/CT-based screening12,13 and one compared usual clinical practice with an extensive screening strategy.4
The results of the risk of bias assessment using the QUADAS-2 and Newcastle-Ottawa Scale tools are provided in Supplementary Tables 1 and 2. No potential sources of bias or
Chapter 10
192
applicability concerns were identified for our review question for the seven studies that had included patients prior to screening procedures. The three other studies were judged to be at high risk of selection bias, given that patients were enrolled after screening.4,12,13 One study was considered to be at high risk of attrition bias due to a considerable proportion of patients lost to follow-up.13 One study was at risk of bias due to inadequate outcome assessment.16 One study that had compared limited with extensive screening was a non-randomized trial.2
Patient-level results
Baseline information
Of the 2,371 patients, 55 (2.3%) had to be excluded from the dataset because they were enrolled more than 90 days after the VTE diagnosis (N=52) or because the index VTE was not confirmed (N=3). Baseline characteristics of the 2,001 patients included in the seven studies that enrolled patients before screening and of all 2,316 patients are shown in Table 3. The mean age was 60 years, 61% were male, and 47% had pulmonary embolism with or without deep vein thrombosis.
Period prevalence of occult cancer
In the seven studies that enrolled patients prior to screening,5,14–18 cancer was diagnosed in 101 of 2,001 patients in the first 12 months, corresponding to a pooled period prevalence of 5.2% (95% CI, 4.1 to 6.5; N=7; Figure 2). When restricting the analysis to the 1,877 patients alive and not lost to follow-up during the first year, the prevalence was 5.7% (95% CI, 4.3 to 7.4; N=7). The different cancer types are summarized in Table 4. Colon cancer was most frequently diagnosed (17%), followed by lung (15%) and pancreatic cancer (11%). In the studies that included patients prior to initial screening,5,14–18 the point prevalence of cancer is shown for different time points in Figure 3. Cancer was detected in 71 of 2,001 patients at initial screening (3.5%; 95% CI, 2.8 to 4.5%; N=7) and diagnosed in 30 of 1,930 patients between the initial screening and the 12-month follow-up (1.6%; 95% CI, 1.0 to 2.6; N=7). The period prevalence of cancer between 12 and 24 months following the index VTE diagnosis was 1.0% (95% CI, 0.56 to 1.9%; N=5). Similarly, in the studies that enrolled patients after initial screening, the period prevalence in the first 12 months of follow-up in patients with negative initial limited screening was 3.5% (95% CI, 1.7 to 6.9; N=10); the period prevalence between 12 and 24 months was 1.1% (95% CI, 0.62 to 1.8; N=8). The sensitivity analysis, restricted to the 2,119 patients enrolled within 30 days of the VTE diagnosis, yielded similar results.
Screening for occult cancer in unprovoked venous thromboembolism
193
10
applicability concerns were identified for our review question for the seven studies that had included patients prior to screening procedures. The three other studies were judged to be at high risk of selection bias, given that patients were enrolled after screening.4,12,13 One study was considered to be at high risk of attrition bias due to a considerable proportion of patients lost to follow-up.13 One study was at risk of bias due to inadequate outcome assessment.16 One study that had compared limited with extensive screening was a non-randomized trial.2
Patient-level results
Baseline information
Of the 2,371 patients, 55 (2.3%) had to be excluded from the dataset because they were enrolled more than 90 days after the VTE diagnosis (N=52) or because the index VTE was not confirmed (N=3). Baseline characteristics of the 2,001 patients included in the seven studies that enrolled patients before screening and of all 2,316 patients are shown in Table 3. The mean age was 60 years, 61% were male, and 47% had pulmonary embolism with or without deep vein thrombosis.
Period prevalence of occult cancer
In the seven studies that enrolled patients prior to screening,5,14–18 cancer was diagnosed in 101 of 2,001 patients in the first 12 months, corresponding to a pooled period prevalence of 5.2% (95% CI, 4.1 to 6.5; N=7; Figure 2). When restricting the analysis to the 1,877 patients alive and not lost to follow-up during the first year, the prevalence was 5.7% (95% CI, 4.3 to 7.4; N=7). The different cancer types are summarized in Table 4. Colon cancer was most frequently diagnosed (17%), followed by lung (15%) and pancreatic cancer (11%). In the studies that included patients prior to initial screening,5,14–18 the point prevalence of cancer is shown for different time points in Figure 3. Cancer was detected in 71 of 2,001 patients at initial screening (3.5%; 95% CI, 2.8 to 4.5%; N=7) and diagnosed in 30 of 1,930 patients between the initial screening and the 12-month follow-up (1.6%; 95% CI, 1.0 to 2.6; N=7). The period prevalence of cancer between 12 and 24 months following the index VTE diagnosis was 1.0% (95% CI, 0.56 to 1.9%; N=5). Similarly, in the studies that enrolled patients after initial screening, the period prevalence in the first 12 months of follow-up in patients with negative initial limited screening was 3.5% (95% CI, 1.7 to 6.9; N=10); the period prevalence between 12 and 24 months was 1.1% (95% CI, 0.62 to 1.8; N=8). The sensitivity analysis, restricted to the 2,119 patients enrolled within 30 days of the VTE diagnosis, yielded similar results.
Table 3. Patient characteristics
Studies with pre‐screening enrolment
(N=2,001)
All studies (N=2,316)
Age, mean (SD), y 58 (15) 60 (15) Age categories ≤49 years, n (%) 592 (30) 628 (27) 50‐74 years, n (%) 1101 (55) 1,263 (55) ≥75 years, n (%) 308 (15) 425 (18) Male, n (%) 1,243 (62) 1,416 (61) Weight, mean (SD), kg 87 (18) 87 (18) Missing, n (%) 640 (32) 955 (41) Index VTE PE with or without DVT, n (%) 950 (47) 1,082 (47) Proximal DVT leg, n (%) 987 (49) 1,132 (49) Distal DVT leg, n (%) 42 (2.1) 50 (2.2) DVT leg with unknown location, n (%) 8 (0.4) 36 (1.6) Upper extremity DVT, n (%) 14 (0.7) 16 (0.7) Evaluation of PE by CTPA, n (%) 750 (38) 750 (32) Unknown, n (%) 1 (0) 121 (5.2) Time from index VTE to enrolment, median (IQR), d 7 (1‐11) 7 (1‐13) Enrolment within 30 days of VTE, n (%) 1,852 (93) 2,119 (92) Smoking Current of former smoker, n (%) 901 (45) 901 (39) Never smoked, n (%) 989 (49) 989 (43) Missing, n (%) 111 (5.5) 426 (18) Previous cancer, n (%) 75 (3.7) 77 (3.3) Missing, n (%) 585 (29) 802 (35) Previous VTE, n (%) 187 (9.3) 192 (8.3) Unprovoked, n (%) 49 (2.4) 49 (2.1) Provoked, n (%) 91 (4.5) 91 (3.9) Missing, n (%) 47 (2.3) 52 (2.2) Estrogen use, n (%) 151 (7.5) 151 (6.5) Missing, n (%) 13 (0.6) 35 (1.5) Screening strategy Limited screening 885 (44) 982 (42) Extensive screening 1,116 (56) 1,334 (58)
Abbreviations: CTPA, computed tomography pulmonary angiography; DVT, deep vein thrombosis; IQR, interquartile range; PE, pulmonary embolism; SD, standard deviation; VTE, venous thromboembolism.
Effect of type of screening
Three studies directly compared a limited to an extensive imaging-based screening strategy.2,3,5 In these studies, the 12-month period prevalence of cancer was 4.2% (95% CI, 3.0 to 5.7%; N=3) and 5.6% (95% CI, 4.3 to 7.3; N=3) among the 885 and 945 patients who underwent the limited or extensive screening strategies, respectively (adjusted odds ratio [OR] 1.4; 95% CI, 0.89 to 2.1; P=0.146).
Chapter 10
194
Figure 2. Period prevalence of cancer in first 12 months of follow‐up Summary period prevalence: 5.2% (95% CI, 4.1 to 6.5; 95% PI, 3.3 to 8.1)
Figure 3. Period prevalence of cancer for different follow‐up periods
Data are based on studies enrolling patients prior to screening procedures. Solid error bars represent the 95% confidence intervals and dashed error bars represent the 95% prediction interval. The y‐axis is truncated at 10%. For these analyses, studies were used that had enrolled patients prior to screening procedures.
Screening for occult cancer in unprovoked venous thromboembolism
195
10
Figure 2. Period prevalence of cancer in first 12 months of follow‐up Summary period prevalence: 5.2% (95% CI, 4.1 to 6.5; 95% PI, 3.3 to 8.1)
Figure 3. Period prevalence of cancer for different follow‐up periods
Data are based on studies enrolling patients prior to screening procedures. Solid error bars represent the 95% confidence intervals and dashed error bars represent the 95% prediction interval. The y‐axis is truncated at 10%. For these analyses, studies were used that had enrolled patients prior to screening procedures.
Table 4. Cancer types and stages
Overall N=2,316a
In 1st year N=2,001b
Beyond 1st year
N=1,624a
Detected by limited
screening tests N=2,001
Detected by
extensive screening
tests N=2,001
Solid cancer 136 (90) 90 (89) 18 (82) 47 (87) 16 (94) Colorectal 26 (19) 17 (16) 7 (39) 7 (15) 5 (31) Lung 21 (15) 15 (14) 1 (5.6) 7 (15) 2 (13) Pancreas 14 (10) 11 (12) 0 (0) 6 (13) 2 (13) Prostate 13 (9.6) 10 (9.4) 2 (11.1) 7 (15) 0 (0) Renal 9 (6.6) 4 (3.8) 1 (5.6) 2 (4.3) 1 (6.2) Breast 7 (5.1) 5 (4.7) 1 (5.6) 4 (8.5) 0 (0) Stomach 7 (5.1) 3 (2.8) 0 (0) 1 (2.1) 1 (6.2) Endometrial 6 (4.4) 5 (4.7) 1 (5.6) 4 (8.5) 0 (0) Bladder 6 (4.4) 2 (1.9) 3 (17) 1 (2.1) 0 (0) Hepatobiliary 6 (4.4) 5 (4.7) 0 (0) 1 (2.1) 3 (19) Unknown primary 3 (2.2) 3 (2.8) 0 (0) 3 (6.4) 0 (0) Esophagus 3 (2.2) 2 (1.9) 0 (0) 1 (2.1) 0 (0) Melanoma 2 (1.5) 2 (1.9) 0 (0) 1 (2.1) 0 (0) Head/neck 2 (1.5) 2 (1.9) 0 (0) 0 (0) 1 (6.2) Ovarian 2 (1.5) 1 (0.9) 0 (0) 1 (2.1) 0 (0) Thyroid 2 (1.5) 0 (0) 0 (0) 0 (0) 0 (0) Testicular 1 (0.7) 1 (0.9) 0 (0) 0 (0) 1 (6.2) Sarcoma 1 (0.7) 0 (0) 1 (5.6) 0 (0) 0 (0) Brain 1 (0.7) 0 (0) 1 (5.6) 0 (0) 0 (0) Adrenal 1 (0.7) 0 (0) 0 (0) 0 (0) 0 (0) Two tumors 3 (2.2) 2 (1.9) 0 (0) 1 (2.1) 0 (0) Hematological cancer 15 (10) 11 (11) 4 (18) 7 (13) 1 (6) Lymphoma 8 (53) 6 (55) 2 (50) 4 (57) 1 (100) Acute leukemia 3 (20) 3 (27) 0 (0) 1 (14) 0 (0) Polycythemia vera 2 (13) 1 (9.1) 1 (25) 1 (14) 0 (0) CLL 1 (6.7) 1 (9.1) 0 (0) 1 (14) 0 (0) Multiple myeloma 1 (6.7) 0 (0) 1 (25) 0 (0) 0 (0) Solid cancer Stage 0 5 (3.7) 4 (4.4) 1 (5.6) 3 (6.4) 0 (0) Stage 1 31 (23) 22 (24) 5 (28) 10 (21) 4 (25) Stage 2 20 (15) 10 (11) 5 (28) 3 (6.4) 4 (25) Stage 3 23 (17) 10 (11) 1 (5.6) 6 (13) 2 (13) Stage 4 56 (41) 43 (48) 6 (33) 25 (53) 5 (31) Unknown 1 (0.7) 1 (1.1) 0 (0) 0 (0) 1 (6.2)
Abbreviations: CLL, chronic lymphocytic leukemia. Data are n (%). a. Data from all ten studies (N=2,316) b. Data from seven studies enrolling patients prior to screening (N=2,001)
Chapter 10
196
Cancer prevalence in subgroups
In the studies that had enrolled patients prior to initial screening,5,14–18 the 12-month period prevalence of cancer was higher in elderly patients, ranging from 0.5% (95% CI, 0.03 to 8.2; N=7) in patients younger than 40 years of age to 9.1% (95% CI, 5.6 to 15; N=7) in those over 80 years (Figure 4). The 12-month period prevalence in predefined subgroups is shown in Table 5. The 12-month period prevalence was 6.8% (95% CI, 5.6 to 8.3; N=7) in patients 50 years or older compared to 1.0% (95% CI, 0.5 to 2.3%) in those younger than 50 years (OR 7.1; 95% CI 3.1 to 16; N=7; P<0.001). The 12-month period prevalence of cancer was 1.3% (95% CI, 0.3 to 5.1; N=4) among women using estrogens and 5.8% (95% CI, 3.8 to 8.8%; N=7) among those not using them.
Figure 4. Point prevalence of cancer at 12 months stratified by age cohorts
Data are based on studies enrolling patients prior to screening procedures. Solid error bars represent the 95% confidence intervals and dashed error bars represent the 95% prediction interval. The y‐axis is truncated at 20%. For these analyses, studies were used that had enrolled patients prior to screening procedures
Cancers detected at screening
Occult cancer was detected at screening in 21 of 885 patients (2.4%; 95% CI, 1.6 to 3.6; N=3) who underwent a limited screening strategy, compared to 50 of 1,116 (4.5%; 95% CI, 3.4 to 5.9%; N=7) in those who underwent extensive screening. The types of cancers detected are shown in Table 4. In the studies that had directly compared a limited to an extensive
Screening for occult cancer in unprovoked venous thromboembolism
197
10
Cancer prevalence in subgroups
In the studies that had enrolled patients prior to initial screening,5,14–18 the 12-month period prevalence of cancer was higher in elderly patients, ranging from 0.5% (95% CI, 0.03 to 8.2; N=7) in patients younger than 40 years of age to 9.1% (95% CI, 5.6 to 15; N=7) in those over 80 years (Figure 4). The 12-month period prevalence in predefined subgroups is shown in Table 5. The 12-month period prevalence was 6.8% (95% CI, 5.6 to 8.3; N=7) in patients 50 years or older compared to 1.0% (95% CI, 0.5 to 2.3%) in those younger than 50 years (OR 7.1; 95% CI 3.1 to 16; N=7; P<0.001). The 12-month period prevalence of cancer was 1.3% (95% CI, 0.3 to 5.1; N=4) among women using estrogens and 5.8% (95% CI, 3.8 to 8.8%; N=7) among those not using them.
Figure 4. Point prevalence of cancer at 12 months stratified by age cohorts
Data are based on studies enrolling patients prior to screening procedures. Solid error bars represent the 95% confidence intervals and dashed error bars represent the 95% prediction interval. The y‐axis is truncated at 20%. For these analyses, studies were used that had enrolled patients prior to screening procedures
Cancers detected at screening
Occult cancer was detected at screening in 21 of 885 patients (2.4%; 95% CI, 1.6 to 3.6; N=3) who underwent a limited screening strategy, compared to 50 of 1,116 (4.5%; 95% CI, 3.4 to 5.9%; N=7) in those who underwent extensive screening. The types of cancers detected are shown in Table 4. In the studies that had directly compared a limited to an extensive
screening strategy,2,3,5 extensive screening was associated with a two-fold higher probability of occult cancer being detected (adjusted OR 2.0; 95% CI, 1.2 to 3.4; P=0.012; N=3) at initial screening. However, additional investigations to pursue cancer diagnosis were performed in 17% of patients undergoing a limited screening strategy, as compared to 26% in those receiving extensive screening (OR 1.7; 95% CI, 0.89 to 3.1; P=0.112). Among patients with a positive limited screening, the probability of cancer being detected was 14% (95% CI, 9.1 to 20%; N=3). The screening tests that were performed in the individual studies are reported in Table 6. Of the 50 cancers detected in patients undergoing an extensive screening strategy, 33 (66%) were detected by limited screening investigations (e.g. medical history). Index screening investigations that led to cancer suspicion and subsequent diagnosis are listed in Table 7. Medical history and physical examination detected 32 of the 71 cancers (45%) diagnosed at initial screening.
Table 5. 12‐month period prevalence of cancer in subgroups
Subgroup Cancer Patients Estimated 12‐month period prevalence % (95% CI)
Age ≥50 years 6 592 6.7 (5.5‐8.2) <50 years 95 1,409 1.0 (0.46‐2.2) Sex Female 38 758 5.0 (3.4‐7.5) Male 63 1,243 5.7 (3.8‐8.5) Estrogen use Yes 2 151 1.3 (0.33‐5.1) No 35 602 5.8 (3.8‐8.8) Smoking Current or former smoker 51 901 5.7 (4.3‐7.4) Never smoked 38 989 3.9 (2.5‐6.0) Index venous thromboembolism Pulmonary embolism ± DVT 42 950 5.2 (3.2‐8.2) DVT only 59 1,051 5.6 (4.4‐7.2) Previous venous thromboembolism
Yes 12 187 6.4 (3.7‐11) No 89 1,814 5.2 (3.8‐7.1)
Abbreviations: CI, confidence interval; DVT, deep vein thrombosis. Analyses are based on studies enrolling patients prior to screening procedures.
Chapter 10
198
Cancer stage
Of the 54 cancers detected by a limited screening strategy, 16 (30%) were stage 0, 1, or 2 solid cancers, 31 (57%) were stage 3 or 4 solid cancer, and 7 (13%) were hematological malignancies (Table 4). Of the 17 cancers detected by an extensive screening strategies, 8 (47%) were stage 0, 1 or 2 solid cancers, 7 (41%) were stage 3 or 4 solid cancer, and 1 (5.9%) was a hematological malignancy, while cancer stage could not be determined in one patient (5.9%). The difference between the proportion of early-stage solid cancers detected by limited (16 of 46) and extensive screening tests (8 of 17) was not statistically significant (P=0.30).
Table 6. Screening tests performed
Limited screening groups (N=885) n (%)
Extensive screening groups (N=1,116)
n (%) Laboratory tests Hemoglobin 883 (99) 1,105 (99) White blood cell count 878 (99) 1,091 (98) Platelet count 877 (99) 1,102 (99) Creatinine 680 (77) 907 (81) Liver function tests ASAT 847 (96) 1041 (93) ALAT 598 (68) 741 (66) Alkaline phosphatase 851 (96) 1,020 (91) Gamma‐glutamyl transferase 167 (19) 223 (20) Bilirubin 599 (68) 697 (62) Calcium 319 (36) 507 (45) Lactate dehydrogenase 665 (75) 786 (70) Prostate specific antigen 287/531 (54) 381/712 (54) Imaging Chest X‐ray 799 (90) 850 (76) US abdomen 1 (0.1) 29 (2.6) Mammography 114/354 (32) 190/404 (47) CT abdomen 4 (0.4) 761 (68) CT chest 3 (0.3) 254 (23) PET/CT 0 (0) 213 (19) Other Pap‐smear 132/354 (37) 125/404 (31)
Screening for occult cancer in unprovoked venous thromboembolism
199
10
Cancer stage
Of the 54 cancers detected by a limited screening strategy, 16 (30%) were stage 0, 1, or 2 solid cancers, 31 (57%) were stage 3 or 4 solid cancer, and 7 (13%) were hematological malignancies (Table 4). Of the 17 cancers detected by an extensive screening strategies, 8 (47%) were stage 0, 1 or 2 solid cancers, 7 (41%) were stage 3 or 4 solid cancer, and 1 (5.9%) was a hematological malignancy, while cancer stage could not be determined in one patient (5.9%). The difference between the proportion of early-stage solid cancers detected by limited (16 of 46) and extensive screening tests (8 of 17) was not statistically significant (P=0.30).
Table 6. Screening tests performed
Limited screening groups (N=885) n (%)
Extensive screening groups (N=1,116)
n (%) Laboratory tests Hemoglobin 883 (99) 1,105 (99) White blood cell count 878 (99) 1,091 (98) Platelet count 877 (99) 1,102 (99) Creatinine 680 (77) 907 (81) Liver function tests ASAT 847 (96) 1041 (93) ALAT 598 (68) 741 (66) Alkaline phosphatase 851 (96) 1,020 (91) Gamma‐glutamyl transferase 167 (19) 223 (20) Bilirubin 599 (68) 697 (62) Calcium 319 (36) 507 (45) Lactate dehydrogenase 665 (75) 786 (70) Prostate specific antigen 287/531 (54) 381/712 (54) Imaging Chest X‐ray 799 (90) 850 (76) US abdomen 1 (0.1) 29 (2.6) Mammography 114/354 (32) 190/404 (47) CT abdomen 4 (0.4) 761 (68) CT chest 3 (0.3) 254 (23) PET/CT 0 (0) 213 (19) Other Pap‐smear 132/354 (37) 125/404 (31)
Table 7. Index tests leading to cancer suspicion and subsequent diagnosis
Limited screening group (N=21) No. (%)
Extensive screening group (N=50) No. (%)
Limited screening tests Medical history / clinical course 12 (57) 14 (28) Blood counts 4 (19) 5 (10) Physical examination 1 (4.8) 5 (10) CT pulmonary angiography 1 (4.8) 5 (10) Prostate specific antigen 3 (14) 3 (14) Mammography 1 (4.8) 4 (8.0) Liver function tests 3 (14) 1 (2.0) Chest X‐ray 3 (14) 1 (2.0) Compression ultrasonography 0 (0) 1 (2.0) Extensive screening tests CT abdomen 0 (0) 10 (20) PET/CT 0 (0) 5 (10) CT chest 0 (0) 1 (2.0) US abdomen 1 (4.8) 0 (0)
Abbreviations: CT, computed tomography; PET, positron emission tomography; US, ultrasonography.
DISCUSSION In this meta-analysis of individual patient data on 2,316 patients with unprovoked VTE, cancer was diagnosed in one of every twenty patients within a year after their VTE diagnosis. About two-thirds of these cancers were detected by screening tests, while the remaining third became clinically overt during follow-up. The probability of a cancer diagnosis was strongly associated with age; the 12-month period prevalence in patients younger than 40 years was negligible, while it was almost 10% in those older than 80 years. An extensive occult cancer screening strategy seems to detect approximately twice as many occult cancers as a more limited strategy, although no statistically significant difference in the proportion of detected early-stage cancers was observed. A few limitations to our analysis should be acknowledged. Despite the availability of patient-level data, there were differences between studies in patient selection, patient characteristics, and screening strategies. This applies especially to the variation in the extensive screening strategies, which complicates the interpretation of the findings, and hampers strong recommendations for clinical practice. There was also variation across the different studies in the time between the index VTE and study enrolment, which could have contributed to the between study variability. Follow-up duration of most studies was 1 to 2 years; hence long-term data on mortality were not available. Yet, this outcome is most important when assessing the potential benefit of screening strategies. Subgroup analysis
Chapter 10
200
based on prior history of cancer diagnosis could not be performed due to a high proportion of missing data (approximately 30%). We are aware that some of the additional analyses are based on small numbers, resulting in wide prediction intervals that reflect the uncertainty around the estimates. However, the main conclusions are based on the collected data of over two thousand VTE patients. We estimated the 12-month period prevalence of occult cancer detection at 5.2%. This is approximately 50% lower than the previously reported 12-month period cumulative incidence from another systematic review, which had included older studies.9 Next to differences in patient populations, there are other potential explanations for this substantial difference. We excluded studies with a retrospective design, given that they may be prone to selection of higher risk patients. A large proportion of the patients in the present analysis were enrolled in large, multicenter trials, which are less prone to overestimation due to single-center bias. VTE practice variation, with more secondary or tertiary care centers in the included centers, may have also resulted in inclusion of a group of patients with lower cancer prevalence. Nonetheless, the rate observed in this analysis is more likely to be representative of the current VTE practice. Hence, this contemporary probability estimate provided in our individual patient data meta-analysis allows clinicians to better inform their patients and assess the risk-benefit ratio of screening procedures. Two other systematic reviews and study-level meta-analyses were identified by searching MEDLINE from inception to May 2017.20,21 In contrast to the present patient-level meta-analysis, these studies solely focused on the potential benefit of extensive screening over limited screening and did not address the period prevalence of cancer overall and in subgroups. The period prevalence of cancer in the second year after the index VTE was estimated at approximately 1% in the present study. This is comparable to the annual cancer incidence observed in the general population, within the same age group.22 This finding is consistent with previously published population-based and cohort studies, in which the cancer probability in VTE patients after six to twelve months of follow-up mirrored that in the general population.23–25 Hence, continued routine surveillance and increased awareness for cancer beyond the first year may not be indicated. Age seems to be an important predictor for the presence of occult cancer in patients with unprovoked VTE. Patients of 50 years of age or older are seven times more likely to be diagnosed with cancer compared to those younger than 50 years. Although physicians often worry about potential cancer in young patients with unprovoked VTE, the probability is actually much higher in older patients. Similarly, the probability of occult cancer appears to
Screening for occult cancer in unprovoked venous thromboembolism
201
10
based on prior history of cancer diagnosis could not be performed due to a high proportion of missing data (approximately 30%). We are aware that some of the additional analyses are based on small numbers, resulting in wide prediction intervals that reflect the uncertainty around the estimates. However, the main conclusions are based on the collected data of over two thousand VTE patients. We estimated the 12-month period prevalence of occult cancer detection at 5.2%. This is approximately 50% lower than the previously reported 12-month period cumulative incidence from another systematic review, which had included older studies.9 Next to differences in patient populations, there are other potential explanations for this substantial difference. We excluded studies with a retrospective design, given that they may be prone to selection of higher risk patients. A large proportion of the patients in the present analysis were enrolled in large, multicenter trials, which are less prone to overestimation due to single-center bias. VTE practice variation, with more secondary or tertiary care centers in the included centers, may have also resulted in inclusion of a group of patients with lower cancer prevalence. Nonetheless, the rate observed in this analysis is more likely to be representative of the current VTE practice. Hence, this contemporary probability estimate provided in our individual patient data meta-analysis allows clinicians to better inform their patients and assess the risk-benefit ratio of screening procedures. Two other systematic reviews and study-level meta-analyses were identified by searching MEDLINE from inception to May 2017.20,21 In contrast to the present patient-level meta-analysis, these studies solely focused on the potential benefit of extensive screening over limited screening and did not address the period prevalence of cancer overall and in subgroups. The period prevalence of cancer in the second year after the index VTE was estimated at approximately 1% in the present study. This is comparable to the annual cancer incidence observed in the general population, within the same age group.22 This finding is consistent with previously published population-based and cohort studies, in which the cancer probability in VTE patients after six to twelve months of follow-up mirrored that in the general population.23–25 Hence, continued routine surveillance and increased awareness for cancer beyond the first year may not be indicated. Age seems to be an important predictor for the presence of occult cancer in patients with unprovoked VTE. Patients of 50 years of age or older are seven times more likely to be diagnosed with cancer compared to those younger than 50 years. Although physicians often worry about potential cancer in young patients with unprovoked VTE, the probability is actually much higher in older patients. Similarly, the probability of occult cancer appears to
be low in women using estrogens. This raises the question whether screening for occult cancer should be offered to low risk patients. Additional diagnostic investigations following an abnormal screening strategy to pursue a cancer diagnosis were frequently performed in studies included in our analysis. Given the costs and potential complications associated with these additional investigations, clinicians may consider foregoing screening tests in these low risk patients. This meta-analysis highlights the fact that the prevalence of occult cancer in patients with unprovoked VTE is substantial. Clinicians should have a low threshold for suspecting cancer in this patient population. A thorough medical history and physical examination are probably the most important screening components, since they led to the diagnosis of more than one-third of cancers at initial screening in the studies included in our analysis. Although a complete blood count and liver function tests appeared to be effective as well, the value of other blood tests (e.g. lactate dehydrogenase and calcium) or Pap-smear seems questionable, given that none of them led to the detection of occult cancers. An extensive screening strategy was associated with a two-fold higher probability of cancer detection at initial screening, at the obvious expense of an increase in the number of targeted tests for cancer. Despite the substantial increase in cancer detection with extensive screening, there is currently not enough evidence to support routine use of these tests in patients with unprovoked VTE. The separate extensive screening tests, such as a CT of the abdomen or whole-body PET/CT, were not associated with a clear increase in detected (early-stage) cancers. The absolute increase of approximately 2% corresponds to a considerable number needed to test, of about 50. Given the relatively short follow-up duration of the included studies, it remains unclear whether the increased cancer detection by extensive screening tests will translate into benefits in patient important outcomes, such as lower morbidity and mortality.
Chapter 10
202
REFERENCES1. Piccioli A, Lensing AWA, Prins MH, Falanga A,
Scannapieco GL, Ieran M, et al. Extensive screening for occult malignant disease in idiopathic venous thromboembolism: a prospective randomized clinical trial. J Thromb Haemost. 2004 Jun;2(6):884–9.
2. van Doormaal FF, Terpstra W, Van Der Griend R, Prins MH, Nijziel MR, Van De Ree MA, et al. Is extensive screening for cancer in idiopathic venous thromboembolism warranted? J Thromb Haemost. 2011 Jan;9(1):79–84.
3. Carrier M, Lazo‐Langner A, Shivakumar S, Tagalakis V, Zarychanski R, Solymoss S, et al. Screening for Occult Cancer in Unprovoked Venous Thromboembolism. N Engl J Med. 2015 Aug 20;373(8):697–704.
4. Prandoni P, Bernardi E, Valle FD, Visonà A, Tropeano PF, Bova C, et al. Extensive Computed Tomography versus Limited Screening for Detection of Occult Cancer in Unprovoked Venous Thromboembolism: A Multicenter, Controlled, Randomized Clinical Trial. Semin Thromb Hemost. 2016 Nov 20;42(8):884–90.
5. Robin P, Le Roux P‐Y, Planquette B, Accassat S, Roy P‐M, Couturaud F, et al. Limited screening with versus without (18)F‐fluorodeoxyglucose PET/CT for occult malignancy in unprovoked venous thromboembolism: an open‐label randomised controlled trial. Lancet Oncol. 2015 Dec 7;2045(15):1–7.
6. Khorana AA, Carrier M, Garcia DA, Lee AYY. Guidance for the prevention and treatment of cancer‐associated venous thromboembolism. J Thromb Thrombolysis. Springer US; 2016;41(1):81–91.
7. van Es N, Le Gal G, Otten H‐M, Robin P, Piccioli A, Lecumberri R, et al. Screening for cancer in patients with unprovoked venous thromboembolism: protocol for a systematic review and individual patient data meta‐analysis. Accept BMJ Open.
8. Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, et al. Preferred Reporting Items for Systematic Review and Meta‐Analyses of individual participant data: the PRISMA‐IPD Statement. JAMA. 2015 Apr 28;313(16):1657–65.
9. Carrier M, Le Gal G, Wells PS, Fergusson D, Ramsay T, Rodger MA. Systematic review: the Trousseau syndrome revisited: should we screen extensively for cancer in patients with venous thromboembolism? Ann Intern Med. 2008 Sep 2;149(5):323–33.
10. Wells G, Shea B, O’Connell D, Peterson J, Welch V, Losos M, et al. The Newcastle‐Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta‐analyses [Internet]. [cited 2017 Jun 1]. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
11. Whiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS‐2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011 Oct 18;155(8):529–36.
12. Beckers MMJ, Verzijlbergen JF, van Buul MMC, Prins MH, Biesma DH. The potential role of positron emission tomography in the detection of occult cancer in 25 patients with venous thromboembolism. Ann Oncol. 2008;19(6):1203–4.
13. Alfonso A, Redondo M, Rubio T, Del Olmo B, Rodríguez‐Wilhelmi P, García‐Velloso MJ, et al. Screening for occult malignancy with FDG‐PET/CT in patients with unprovoked venous thromboembolism. Int J Cancer. 2013 Apr 24;133(9):1–8.
14. Jara‐Palomares L, Rodríguez‐Matute C, Elías‐Hernández T, Rodríguez‐Portal JA, López‐Campos JL, Garcia‐Ibarra H, et al. Testing for occult cancer in patients with pulmonary embolism: results from a screening program and a two‐year follow‐up survey. Thromb Res. 2010 Jan;125(1):29–33.
15. Carrier M, Le Gal G, Tao H, Wells PS, Danovitch K, Mbanga‐levac A, et al. Should we screen patients with unprovoked venous thromboembolism for occult cancers? A pilot study. Blood Coagul Fibrinolysis. 2010 Oct;21(7):709–10.
16. Rieu V, Chanier S, Philippe P, Ruivard M. Systematic screening for occult cancer in elderly patients with venous thromboembolism: a prospective study. Intern Med J. 2011 Nov;41(11):769–75.
17. Rondina MT, Wanner N, Pendleton RC, Kraiss LW, Vinik R, Zimmerman G a, et al. A pilot study utilizing whole body 18 F‐FDG‐PET/CT as a comprehensive screening strategy for occult malignancy in patients with unprovoked venous thromboembolism. Thromb Res; 2012 Jan;129(1):22–7.
18. Ihaddadene R, Corsi DJ, Lazo‐Langner A, Shivakumar S, Zarychanski R, Tagalakis V, et al. Risk factors predictive of occult cancer detection in patients with unprovoked venous thromboembolism. Blood. 2016;1–13.
Screening for occult cancer in unprovoked venous thromboembolism
203
10
REFERENCES1. Piccioli A, Lensing AWA, Prins MH, Falanga A,
Scannapieco GL, Ieran M, et al. Extensive screening for occult malignant disease in idiopathic venous thromboembolism: a prospective randomized clinical trial. J Thromb Haemost. 2004 Jun;2(6):884–9.
2. van Doormaal FF, Terpstra W, Van Der Griend R, Prins MH, Nijziel MR, Van De Ree MA, et al. Is extensive screening for cancer in idiopathic venous thromboembolism warranted? J Thromb Haemost. 2011 Jan;9(1):79–84.
3. Carrier M, Lazo‐Langner A, Shivakumar S, Tagalakis V, Zarychanski R, Solymoss S, et al. Screening for Occult Cancer in Unprovoked Venous Thromboembolism. N Engl J Med. 2015 Aug 20;373(8):697–704.
4. Prandoni P, Bernardi E, Valle FD, Visonà A, Tropeano PF, Bova C, et al. Extensive Computed Tomography versus Limited Screening for Detection of Occult Cancer in Unprovoked Venous Thromboembolism: A Multicenter, Controlled, Randomized Clinical Trial. Semin Thromb Hemost. 2016 Nov 20;42(8):884–90.
5. Robin P, Le Roux P‐Y, Planquette B, Accassat S, Roy P‐M, Couturaud F, et al. Limited screening with versus without (18)F‐fluorodeoxyglucose PET/CT for occult malignancy in unprovoked venous thromboembolism: an open‐label randomised controlled trial. Lancet Oncol. 2015 Dec 7;2045(15):1–7.
6. Khorana AA, Carrier M, Garcia DA, Lee AYY. Guidance for the prevention and treatment of cancer‐associated venous thromboembolism. J Thromb Thrombolysis. Springer US; 2016;41(1):81–91.
7. van Es N, Le Gal G, Otten H‐M, Robin P, Piccioli A, Lecumberri R, et al. Screening for cancer in patients with unprovoked venous thromboembolism: protocol for a systematic review and individual patient data meta‐analysis. Accept BMJ Open.
8. Stewart LA, Clarke M, Rovers M, Riley RD, Simmonds M, Stewart G, et al. Preferred Reporting Items for Systematic Review and Meta‐Analyses of individual participant data: the PRISMA‐IPD Statement. JAMA. 2015 Apr 28;313(16):1657–65.
9. Carrier M, Le Gal G, Wells PS, Fergusson D, Ramsay T, Rodger MA. Systematic review: the Trousseau syndrome revisited: should we screen extensively for cancer in patients with venous thromboembolism? Ann Intern Med. 2008 Sep 2;149(5):323–33.
10. Wells G, Shea B, O’Connell D, Peterson J, Welch V, Losos M, et al. The Newcastle‐Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta‐analyses [Internet]. [cited 2017 Jun 1]. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
11. Whiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS‐2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011 Oct 18;155(8):529–36.
12. Beckers MMJ, Verzijlbergen JF, van Buul MMC, Prins MH, Biesma DH. The potential role of positron emission tomography in the detection of occult cancer in 25 patients with venous thromboembolism. Ann Oncol. 2008;19(6):1203–4.
13. Alfonso A, Redondo M, Rubio T, Del Olmo B, Rodríguez‐Wilhelmi P, García‐Velloso MJ, et al. Screening for occult malignancy with FDG‐PET/CT in patients with unprovoked venous thromboembolism. Int J Cancer. 2013 Apr 24;133(9):1–8.
14. Jara‐Palomares L, Rodríguez‐Matute C, Elías‐Hernández T, Rodríguez‐Portal JA, López‐Campos JL, Garcia‐Ibarra H, et al. Testing for occult cancer in patients with pulmonary embolism: results from a screening program and a two‐year follow‐up survey. Thromb Res. 2010 Jan;125(1):29–33.
15. Carrier M, Le Gal G, Tao H, Wells PS, Danovitch K, Mbanga‐levac A, et al. Should we screen patients with unprovoked venous thromboembolism for occult cancers? A pilot study. Blood Coagul Fibrinolysis. 2010 Oct;21(7):709–10.
16. Rieu V, Chanier S, Philippe P, Ruivard M. Systematic screening for occult cancer in elderly patients with venous thromboembolism: a prospective study. Intern Med J. 2011 Nov;41(11):769–75.
17. Rondina MT, Wanner N, Pendleton RC, Kraiss LW, Vinik R, Zimmerman G a, et al. A pilot study utilizing whole body 18 F‐FDG‐PET/CT as a comprehensive screening strategy for occult malignancy in patients with unprovoked venous thromboembolism. Thromb Res; 2012 Jan;129(1):22–7.
18. Ihaddadene R, Corsi DJ, Lazo‐Langner A, Shivakumar S, Zarychanski R, Tagalakis V, et al. Risk factors predictive of occult cancer detection in patients with unprovoked venous thromboembolism. Blood. 2016;1–13.
19. Wells PS, Prins MH, Levitan B, Beyer‐Westendorf J, Brighton TA, Bounameaux H, et al. Long‐term Anticoagulation With Rivaroxaban for Preventing Recurrent VTE: A Benefit‐Risk Analysis of EINSTEIN‐Extension. Chest. 2016 Nov;150(5):1059–68.
20. Klein A, Shepshelovich D, Spectre G, Goldvaser H, Raanani P, Gafter‐Gvili A. Screening for occult cancer in idiopathic venous thromboembolism — Systemic review and meta‐analysis. Eur J Intern Med. 2017.
21. Robertson L, Yeoh SE, Stansby G, Agarwal R, Yeoh SE. Effect of testing for cancer on cancer‐ and venous thromboembolism (VTE)‐related mortality and morbidity in patients with unprovoked VTE. Robertson L, editor. Cochrane Database Syst Rev; 2013 Mar 6;(3).
22. Surveillance, Epidemiology, and End Results Program (SEER) age‐adjusted cancer incidence [Internet]. [cited 2017 Mar 6]. Available from: https://seer.cancer.gov/faststats/selections.php?run=runit&output=2&data=1&statistic=1&cancer=1&year=201603&race=1&sex=1&series=age&age=141
23. Sorensen HT, Mellemkjaer L, Steffensen FH, Olsen JH, Nielsen GL, Sørensen HT, et al. The risk of a diagnosis of cancer after primary deep venous thrombosis or pulmonary embolism. N Engl J Med. 1998 Apr 23;338(17):1169–73.
24. Prandoni P, Casiglia E, Piccioli A, Ghirarduzzi A, Pengo V, Gu C, et al. The risk of cancer in patients with venous thromboembolism does not exceed that expected in the general population after the first 6 months. J Thromb Haemost. 2010;8(5):1126–7.
25. Nordström M, Lindblad B, Anderson H, Bergqvist D, Kjellström T. Deep venous thrombosis and occult malignancy: an epidemiological study. BMJ.1994 Apr 2;308(6933):891–4.
Chapter 10
204
Supp
lemen
tary Tab
le 1. Sum
mary of risk of b
ias a
ssessm
ent w
ith QUA
DAS‐2
Risk of b
ias
Applica
bility concerns
STUD
Y Pa
tient
selection
Inde
x test
Reference
stan
dard
Flow
and
tim
ing
Patie
nt
selection
Inde
x test
Reference stan
dard
Alfonso et al.
Beckers e
t al.
?
? ?
?
Carrier e
t al. (201
5)
Jara‐Palom
ares et a
l.
Robin et al.
Pran
doni et a
l.
Rieu
et a
l.
Rond
ina et al.
Carrier e
t al. (201
0)
Van Do
ormaal et a
l.
Supp
lemen
tary Tab
le 2. Sum
mary of risk of b
ias a
ssessm
ent w
ith New
castle‐Ottaw
a Scale
Se
lection
Compa
rability
of Coh
orts
Outcom
e To
tal
Re
presen
tativ
eness
of Exp
osed
Coh
ort
Represen
tativ
eness
of Non
expo
sed
Coho
rt
Ascertainm
ent
of Exp
osure
Outcom
e no
t presen
t at
start study
As
sessmen
t of Outcome
Follo
w‐
up Lo
ng
Enou
gh?
Adeq
uacy
Follo
w‐up
Alfonso et al.
‐ NA
*
* NA
*
* ‐
5 Be
ckers e
t al.
‐ NA
*
* NA
‐
* ‐
3 Ca
rrier e
t al. (201
5)
* *
* *
**
* *
* 9
Jara‐Palom
ares et a
l.
* NA
*
* NA
*
* *
5 Ro
bin et al.
* *
* *
**
* *
* 9
Pran
doni et a
l.
‐ ‐
* *
**
* *
* 9
Rieu
et a
l.
* NA
*
* NA
0
* *
5 Ro
ndina et al.
* NA
*
* NA
*
* *
6 Ca
rrier e
t al. (201
0)
* NA
*
* NA
*
* *
6 Va
n Do
ormaal et a
l.
* *
* *
**
* *
* 9