Venous Thromboembolism in Cancer Patients Undergoing Chemotherapy: A Systematic Review and Meta-Analysis

Objective: Venous thromboembolism (VTE) is a life-threatening complication that may exacerbate cancer prognosis. Whilst some studies indicate an increased risk of VTE in cancer patients undergoing chemotherapy, the prevalence estimates on the pooled prevalence of VTE in cancer patients undergoing chemotherapy are not known. This study aims to calculate the pooled prevalence of VTE in chemotherapy-treated cancer patients. Methods: Studies on VTE occurrence in cancer patients undergoing chemotherapy were retrieved after database search. The terms used included “cancer”, “chemotherapy”, and “venous thromboembolism”. A random-effects meta-analysis was conducted to obtain a pooled estimate of VTE prevalence in cancer patients undergoing chemotherapy. Results: A total of 102 eligible studies involving 30,671 patients (1773 with VTE, 28,898 without) were included in the meta-analysis. The pooled estimate of VTE prevalence was found to be 6%, ranging from 6% to 7% (ES 6%; 95% CI 6–7%; z = 18.53; p < 0.001). Conclusions: The estimated pooled prevalence rate of VTEs was 6% in cancer patients undergoing CRT, which was higher than the overall crude prevalence rate (5.78%). Comprehensive cancer care should consider stratified VTE risk assessment based on cancer phenotype, given that certain phenotypes of cancer such as bladder, gastric and ovarian posing particularly high risks of VTE.


Introduction
Venous thromboembolism (VTE) is a major public health problem constituting a significant burden of disease [1,2]. There are around 10 million cases of VTE worldwide every year. After myocardial infarction and stroke, VTE is the third leading vascular disease [3]. In the first one to three months following a stroke, there is an increased risk of VTE, partly because of immobility brought on by the stroke [4]. Major venous and arterial thrombotic disorders share overlap in some key cardiovascular risk factors [5]. A higher risk of VTE is linked to specific cardiovascular risk factors such as older age, smoking, and greater adiposity [2,6]. Cancer, a leading cause of death and disability in the world [7,8], is known to potentiate the risk of VTE and roughly 20% of VTE are linked to cancer [9,10]. Thrombosis in cancer patients is a clinically challenging construct which is associated with poor outcomes despite therapy [11].
Recent studies have also indicated that cancer patients on chemotherapy may be at an increased risk of venous thromboembolism [12][13][14]. Assessment of VTE risk is critical for appropriate medical management and prophylactic treatment [15]. Given the lack of data on the prevalence estimates of VTE in cancer patients, especially in those receiving chemotherapy, further studies are required. Distinct cancer phenotypes may render cancer patients at varying levels of VTE risk [16][17][18]. Recent guidelines from American Society of Haematology published in early 2021 recommend stratifying cancer patients according to their VTE risk prior to the start of chemotherapy, as well as patientspecific factors, using the Khorana risk score, the major determinant of which is cancer phenotype [19]. This comes in the background of two landmark randomized clinical trials (RCTs), resulting in the change of guidelines, demonstrating VTE prophylaxis with direct oral anticoagulants (DOACs) following risk assessment lowered the incidence of VTE during chemotherapy [20][21][22]. Several societies or health systems beyond United States are yet to adopt these recommendations; besides, unwarranted variations in clinical care as well as poor adherence to recommendations or guidance vis à vis VTE risk assessment and optimal administration of thromboprophylaxis pose an ongoing real-world or systems challenge [23,24]. Moreover, literature is sparse when comparing the relative risk and prevalence of VTE across multiple cancer phenotypes-with studies only revealing VTE prevalence specific to a cancer phenotype and risk in homogenous cancer populations, vis à vis their ethnicity and treatment received. Understanding of, and estimates of, the pooled prevalence may also be useful to increase awareness on VTE risks in cancer patients undergoing chemotherapy as well to inform clinicians and patients on the quantum of the VTE prevalence/risks in cancer or across various types of cancer. This meta-analysis sought to investigate the pooled prevalence of venous thromboembolism in cancer patients receiving chemotherapy. There is also a gap in clinician knowledge pertaining to the specific risk that cancer phenotypes and chemotherapy poses to cancer patients. We have sought to address two key underlying questions through this meta-analysis: (1) what is the prevalence of VTE in cancer patients receiving chemotherapy? (2) what is the prevalence of VTE stratified by cancer phenotype in patients undergoing chemotherapy?

Literature Search: Identification and Selection of Studies
The primary search engine of this meta-analysis and systematic review was the PubMed database. Articles published between 2012 and October 2022 were included in the search. Search terms included: "cancer", "chemotherapy" and "venous thromboembolism". The complete search strategy is available in the Supplementary Information (Search Strategy). Studies were filtered to include those in the English language, conducted on humans, and restricted to disregard Phase I studies, accepting only those Phase II and above. Additional studies were also included through handsearching of references from included studies as well as from other sources such as Google Scholar and ResearchGate. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. This study was registered in Open Science, registration number is "yn5br" (https://osf.io/yn5br/ (accessed on 6 November 2022)). The PRISMA flowchart shows the studies included in the meta-analysis ( Figure 1). PRISMA checklist is also provided in the Supplementary Information (PRISMA Checklist).

Inclusion and Exclusion Criteria
Studies were eligible for inclusion if they met the following criteria: (1) age ≥ 18 years; (2) patients with a confirmed diagnosis of cancer; (3) patients receiving chemotherapy; (4) patients not on prophylactic anticoagulation concomitant to chemotherapy, (5) availability of data on VTE occurrence noted in patients; and (6) studies with a sample size of >20 patients. The exclusion criteria were (1) studies not in English, (2) animal studies, (3) duplicated publications, (4) systematic reviews, meta-analyses, or narrative reviews; and (5) studies whereby relevant data on VTE occurrence not available.

Data Extraction
Firstly, titles and abstracts were screened on EndNote™ (Clarivate, Philadelphia, PA, USA) to identify articles that were beyond the scope of this study, were systematic reviews of meta-analyses, or for other reasons failed to match the eligibility criteria before being excluded. Remaining articles were read in full-text and comprehensively assessed to determine eligibility for inclusion in this study, with screening conducted independently by two experienced investigators. In the case of disagreement between authors, a consensus was reached through discussion. A data extraction sheet was used to extract the following data from each study: (1) baseline demographics: author, year of publication, type of publication, country of lead author, study design, and study type; (2) study population: age of patients, sample size, baseline clinical characteristics, cancer phenotype, body location of cancer, cancer stage, and treatment agent, dose, duration, and frequency; (3) outcome measures: VTE occurrence. In the grading of VTE severity, most VTEs were classified as an adverse effect within a drug trial, and thus were graded via the Common Terminology Criteria for Adverse Events (CTCAE) scale of Grade 1 through 5. Although some studies were particular in grading each adverse event into individual categories of Grade 1/2/3/4/5, the majority of studies grouped Grades 1 and 2 together, and Grades 3/4/5 together. As such, in our meta-analysis, we have extracted data based on this later, more generalised method. Studies reporting on VTE in cancer, without prophylactic anticoagulation concomitant to chemotherapy, were included in the systematic review and meta-analysis.

Quality Assessment of Included Studies
Using the modified Jadad analysis (MJA) criterion, the methodological quality of each study was assessed [25]. The MJA evaluates the quality of studies based upon: randomisation, blinding, description of withdrawals/dropouts, inclusion/exclusion criteria, assessment of adverse events and methods used for statistical analysis. Studies receive a score from 0-8 based upon their ability to fulfil aforementioned criteria [26]. The complete quality assessment of each study is available in the Supplementary Information (Jadad Analysis). Each study was also separately assessed for risk of funding bias using a 2-point scale that scored studies from 0 (low potential for bias) to 2 (high potential for bias) [27]. The absence of industry funding was not taken to signify an absence of bias, but the presence of industry funding or conflicts of interest was assumed to be an indicator of bias.

Statistical Analysis
Statistical analysis was performed using STATA (Version 13.0, StataCorp LLC, College Station, TX, USA). The purpose of this study was to determine the prevalence of VTE in cancer patients undergoing chemotherapy. As a result, the "metaprop" STATA command was utilised, pooling prevalence by performing a random-effects meta-analysis of proportions obtained from the individual studies [28]. The DerSimonian and Laird method was used for random effects modelling. In presenting the overall effects, forest plots were generated. Heterogeneity across the studies was estimated from the inverse-variance fixedeffect model and quantified using the I 2 measure (I 2 < 40% = low, 30-60% = moderate, 50-90% = substantial, and 75-100% = considerable). An overall meta-analysis was performed stratified by cancer phenotype to estimate the pooled prevalence of VTE in cancer patients undergoing chemotherapy. Besides, meta-analysis for individual cancer pheno-types were also performed provided there were minimum of 4 studies. The estimate of between-study variance (tau-squared or τ 2 ) was also reported. Significance tests in the form of z-statistics and p-values were also reported. p-values less than 0.05 were considered significant.

Results
A total of 2643 and 85 studies were identified from PubMed and other sources, respectively. On screening of 2723 titles/abstracts, 2172 studies were excluded. Out of these, full texts of 551 studies were assessed for eligibility. Overall, 102 studies were included in the final synthesis.

Description of Included Studies
This meta-analysis included 102 studies, which reported on VTE prevalence within cancer patients undergoing chemotherapy, with a cumulative cohort of 30,671 patients (1773 with VTE, 28,898 without). Within a cohort of 20,420 patients for which data on sex was reported, 53.11% were male (sex data on 10,251 patients were not reported). Age ranged from 18-93 years. The mean age within a cohort of 8159 patients on which age data was reported was 59.59 years (standard deviation (SD) 33.74). Twenty-two cancer phenotypes were identified including bladder, blood, brain, breast, cervical, colorectal, endometrial, gastric, germ cell, head and neck, liver, lung, lymph, mesothelial, mixed, neuroendocrine, oesophageal, ovarian, pancreatic, prostate, renal, and skin.
The clinical characteristics of all included studies are shown in Table 1. Tables 2 and 3 further delineate treatment dose, duration, and frequency. Tabulated results for the estimated pooled prevalence and crude prevalence rates stratified by cancer phenotype are shown in Table 4. The MJA and funding bias analysis for each study can be found in Supplemental SI Tables S1 and S2.        Patients received one of three docetaxel-based neoadjuvant regimens for four cycles: docetaxel alone (100 mg/m 2 ) with addition of capecitabine (825 mg/m 2 ) oral twice daily days 1-14, 75 mg/m 2 ) docetaxel) or with addition of gemcitabine (1000 mg/m 2 ) days 1 and 8 intravenously, 75 mg/m 2 docetaxel), all followed by neoadjuvant doxorubicin and cyclophosphamide (60 mg/m 2 ) and 600 mg/m 2 ) intravenously) every 3 weeks for four cycles.
Those randomly assigned to bevacizumab groups were to receive bevacizumab (15 mg/kg, every 3 weeks for six cycles) with neoadjuvant chemotherapy and postoperatively for ten doses.    In the mFOLFOX6 and the aflibercept/mFOLFOX6 arms, the median duration of exposure to oxaliplatin was 23.2 (2-77) and 22.0 (2-84), and to 5-FU was 25.9 (2-95) and 24         Thalidomide was given at a dose ranging from 100 mg/day to 400 mg/day continuously; lenalidomide dose was 25 mg/day, days 1 to 21 on a 28-day cycle. All patients received dexamethasone, either at high dose (40 mg orally on days 1-4, 9-12, and 17-20)   Patients received intravenous gemcitabine 1000 mg/m 2 on days 1 and 8 every 3 weeks, combined with either intravenous ramucirumab 10 mg/kg or matching placebo on day 1 of a 3-week cycle, until progressive disease, unacceptable toxicity, or withdrawal of consent to treatment occurred.

Overall Prevalence of VTE in Cancer Patients Undergoing Chemotherapy
One hundred and two included studies, encompassing 30,671 patients, reported on the prevalence of VTE in cancer patients receiving chemotherapy only. The metaanalysis revealed a pooled estimated prevalence of 6%, ranging from 6% to 7% (ES 6%; 95% CI 6-7%; z = 18.53; p < 0.001) ( Table 4 and Figure 2). Notably, there was considerable heterogeneity between the included studies (I 2 = 91.84%, p < 0.001). The estimate of betweenstudy variance (τ 2 ) was 0.04. The estimated pooled prevalence of VTE in cancer patients undergoing chemotherapy was higher than the crude prevalence rates of 5.78% observed in this study. The heterogeneity chi 2 was 1237.89 (p < 0.001, d.f. = 101). Figure 3 provides findings of the overall meta-analysis on the pooled estimated prevalence stratified by cancer phenotype.       . Abbreviations: VTE: venous thromboembolism. Figure 3. Forest plot showing the estimated pooled prevalence of VTE in cancer patients stratified by cancer phenotype . Abbreviations: VTE: venous thromboembolism.

Prevalence of VTE in Blood Cancer Patients
Three included studies, encompassing 934 patients, reported on the prevalence of VTE in blood cancer patients [85,107,117]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in blood cancer patients was 10.81% (Table 4).

Prevalence of VTE in Cervical Cancer Patients
Two included studies, encompassing 716 patients, reported on the prevalence of VTE in cervical cancer patients [63,124]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in cervical cancer patients was 6.42% (Table 4).

Prevalence of VTE in Endometrial Cancer Patients
Three included studies, encompassing 173 patients, reported on the prevalence of VTE in endometrial cancer patients [31,54,74]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in endometrial cancer patients was 11.56% (Table 4).

Prevalence of VTE in Germ Cell Cancer Patients
One included study, encompassing 193 patients, reported on the prevalence of VTE in endometrial cancer patients [65]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in germ cell cancer patients was 2.07% (Table 4).

Prevalence of VTE in Head and Neck Cancer Patients
Two included studies, encompassing 158 patients, reported on the prevalence of VTE in head and neck cancer patients [38,62]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in head and neck cancer patients was 1.27% (Table 4).

Prevalence of VTE in Liver Cancer Patients
Two included studies, encompassing 347 patients, reported on the prevalence of VTE in liver cancer patients [43,109]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in liver cancer patients was 5.19% (Table 4).

Prevalence of VTE in Neuroendocrine Cancer Patients
One included study, encompassing 113 patients, reported on the prevalence of VTE in neuroendocrine cancer patients [56]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in neuroendocrine cancer patients was 6.19% (Table 4).

Prevalence of VTE in Oesophageal Cancer Patients
Two included studies, encompassing 328 patients, reported on the prevalence of VTE in oesophageal cancer patients [52,67]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in oesophageal cancer patients was 9.76% (Table 4).

Prevalence of VTE in Pancreatic Cancer Patients
Three included studies, encompassing 144 patients, reported on the prevalence of VTE in pancreatic cancer patients [32,33,81]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in pancreatic cancer patients was 28.47% (Table 4).

Prevalence of VTE in Prostate Cancer Patients
Three included studies, encompassing 1233 patients, reported on the prevalence of VTE in prostate cancer patients [95,100,108]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in prostate cancer patients was 2.11% (Table 4).

Prevalence of VTE in Renal Cancer Patients
Two included studies, encompassing 198 patients, reported on the prevalence of VTE in renal cancer patients [48,105]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in renal cancer patients was 11.11% (Table 4).

Prevalence of VTE in Skin Cancer Patients
One included study, encompassing 93 patients, reported on the prevalence of VTE in skin cancer patients [75]. However, a meta-analysis could not be performed due to insufficient number of studies. The crude prevalence rate of VTE in skin cancer patients was 7.53% (Table 4).

Discussion
Our meta-analysis revealed an overall pooled estimated prevalence of VTEs in cancer patients undergoing chemotherapy, as well for various cancer phenotypes. Our findings indicate that the estimated pooled prevalence of VTEs in cancer patients undergoing chemotherapy is approximately 6%, ranging from 5% to 7%, which is higher than the crude prevalence rate of 5.78%. To the best of our knowledge, this is one of the first reports in which prevalence estimates of VTE have been conducted on a relatively large cohort of patients. Our findings also reveal phenotypic variability in VTE risk, indicating need for prophylactic management of VTE risk in cancer patients undergoing chemotherapy, with certain phenotypes of cancer such as bladder, gastric and ovarian posing particularly high risks of VTE.
One explanation for why cancer patients have a higher risk of having VTE is that tumours can express various procoagulant molecules and alter tissue factor expression [131,132]. Certain tumours may also raise the risk of thrombosis by compressing blood vessels, changing blood flow, or causing injury to the vascular endothelium through intravascular growth [9]. Subsequent cancer diagnosis within the first year of first VTE diagnosis have been reported in up to 10% of patients [133]. Therefore, VTE, especially in the lower limbs, can also be useful as marker for occult cancer [134].
This pooled estimate of 6% is higher than other estimates of 2.3% prevalence rates of VTE in cancer patients in the first 12 months after their diagnosis, with other estimates ranging from 4-20% of cancer patients developing VTE in their lifetime [12,135]. Amongst the normal population, VTE prevalence is at 1-2% [136]. In a retrospective study on 40,787,000 hospitalised cancer patients from 1979 through 1999, patients with malignancy were found to have a 2% prevalence of thromboembolism, although, were not necessarily on chemotherapy or radiotherapy treatment [137]. This suggests that cancer itself, without the interference of external treatment regimens, may not pose a significant risk to VTE but rather, it is the accompanying therapies which may confer additional VTE risk.
The prevalence of VTE may vary across cancer phenotype. This is of clinical interest as it may aid in the risk-staging and appropriate tailored management specific to cancer phenotype. We found that the pooled VTE prevalence varied across cancer phenotype in the range of 1-18%, with lowest prevalence of 1% observed in breast and head and neck cancer and highest prevalence of 18% observed in bladder cancer. This indicates a need for more aggressive VTE screening for specific cancer phenotypes. From a policy standpoint, beyond the hospital-based risk factors, such as recent surgery, cancer, and congestive heart failure, to prevent VTE, dietary counselling as well as public health strategies around encouraging the adoption of heart-healthy habits for cancer patients undergoing chemotherapy may be beneficial [138]. Moreover, concomitant preventative measures targeting arterial thrombosis and VTE are also important [2].

Pathophysiology of VTE in Cancer Patients
The pathophysiological process behind VTE prevalence in cancer patients is multifaceted and can be attributed to multiple aetiological pathways, spanning the hypercoagulable state induced by malignancy itself to the thrombotic risk posed by treatment regimens of chemotherapy and radiotherapy [139]. The inflammatory state induced by malignancy, stemming from tumour biology and activation of the coagulation cascade, increases cancer patients' risk of VTE occurrence [139]. On a molecular level, several factors increase the risk of VTE, with increased concentrations of procoagulants on a cellular level amplifying thrombosis prevalence. These include tissue factor, microparticles, plasminogen activator inhibitor-1, cancer procoagulant, mucin, tumour-derived platelet agonists and inflammatory cytokines such as IL-6, IL-8 and IL-10 [140][141][142][143][144][145]. Alterations to thrombomod-ulin expression due to interference from tumour necrosis factor-a and IL-1B also contribute to a prothrombotic state [146].

Chemotherapy and VTE
Chemotherapy has been shown to increase VTE risk by six-fold in cancer patients [12]. Multiple chemotherapy drugs which are used to treat cancer are associated with increased thrombotic events [147]. Cisplatin is a major component of several treatment regimensand its thrombotic potential and vascular toxicity has been identified since 1986 [148]. Through direct drug-induced damage to the endothelium and by indirectly increasing the expression of TF procoagulant activity of monocytes and macrophages, chemotherapy poses a serious risk of increasing VTE within a cancer patient [148].

Limitations
There are several limitations to this study due to variance across the quality of studies included and therefore ability to accurately process the data extracted. Firstly, the types of studies included vary from being retrospective in design to being randomised controlled trials (RCT)-therefore, whilst some studies noted VTE as one of multiple adverse effects within an RCT for a novel chemotherapy regimen, others purely sought to document VTE occurrence within a cohort of cancer patients which oftentimes varied in cancer phenotype, staging and treatment. Furthermore, whilst some studies were robust in being double-blinded, randomised, placebo-controlled and multi-institutional, others were single institution studies conducted on a relatively small cohort size, without an appropriate control group or blinding. As such, this wide variance in included study quality could confound the overall pooled estimated prevalence. Within these studies, their documentation of patients' cancer history is highly limited. Reporting of time of diagnosis to treatment, the duration, drug regimen and frequency of previous treatments are inconsistent and rarely available. As such, it is difficult to determine whether previous treatment regimens played a confounding role in patients developing VTE. We are also unable to determine whether variance in time of diagnosis to treatment plays a role. Moreover, there was a lack of standardised reporting and insufficient detail in the description of VTEs across the studies. As the majority of studies included were RCTs, VTEs were often a side effect as opposed to the focus of the study, and thus less attention was given toward the VTE. VTE pooled prevalence stratified by cancer severity grade was not investigated in this meta-analysis. In few cases, studies neglected to document the severity of thrombotic events altogether-in which we have assumed a Grade 3/4/5 event in that case. Besides, detailed analysis into the association between severity of VTE event, and any relationship with cancer phenotype, time to treatment, staging, drug regimen, or patient profile could not be performed. Additionally, documentation on certain groups such as atrial fibrillation and VTE recurrence were not available across all studies. It would be ideal to understand whether VTE events occurred before therapy, during, or how long post-diagnosis and post-treatment. As such, more robust future studies with more detailed information and reporting on VTE occurrence, recurrence and adverse effects is necessary. Another limitation of this study and in the studies gathered is the lack of accounting for baseline underlying comorbidities in all the patients. Important factors such as atherosclerosis, cardiovascular disease, histories of smoking, histories of VTE, obesity and age were not detailed in the original studies.
Although patients on prophylactic anticoagulation concomitant to chemotherapy were not included in this study, we acknowledge that previous history of anticoagulation may presumably not have been reported in some studies. In light of recent guidelines [19], as adherence to prophylactic anticoagulation grows to reduce VTE risk, it is likely that VTE prevalence will show a downward trajectory. Finally, for the patients who did experience VTE prevalence, often these patients were not followed longitudinally for VTE recurrence, and specific time to disease progression and overall survival. Typically, follow-up was not provided for longer-term complications and recurrence. The discrepancy in protocol for VTE diagnosis and follow-up between hospitals and studies leads to the inconsistent reporting and treatment of patients across the clinical decision making, imaging and diagnosis pipeline. Despite these limitations, the use of random-effects modelling would have mitigated some of the random biases and risks above.

Conclusions
In conclusion, this meta-analysis demonstrated a pooled prevalence estimate of 6%, with a range of 5% to 7%, of VTEs amongst cancer patients undergoing chemotherapy. Our study indicates there is substantial risk of developing VTE as a cancer patient on chemotherapy showing a compelling need for robust screening and subsequent prophylactic management to prevent future VTE. More efforts should be undertaken to implement adherence of American Society of Haematology guidelines on VTE risks and management in cancer patients undergoing chemotherapy [19].  Institutional Review Board Statement: Ethical review and approval is not applicable for this study as all analyses were based on previously published studies; thus, no ethical approval or patient consent was required.

Informed Consent Statement: Not applicable.
Data Availability Statement: The original contributions presented in the study are included in the article, and further inquiries can be directed to the corresponding author.