Abstract
Background
Venous thromboembolism (VTE) is an umbrella term for the pathological development of blood clots in the venous system. These include pulmonary embolism (PE), deep and superficial venous thromboses. A cancer patient’s risk of developing VTE is at least 4-fold that of the general population and can increase markedly when undertaking chemotherapy. With advances in radiological imaging, increasing numbers of clinically unsuspected VTEs are diagnosed on routine cancer staging scans. This may be because the patient is asymptomatic, the clinician may be unaware of the increased risk or the non-specific presenting symptoms and signs may be attributed to the underlying cancer. VTE is ranked as the second biggest killer of cancer inpatients and a leading killer of outpatients worldwide.
Study aims
1. Establish the prevalence/cumulative incidence of venous thromboembolism (VTE) in cancer patients undertaking chemotherapy at the Canterbury Regional Cancer and Haematology Service (CRCHS).
2. Identify potential clinicopathological biomarkers predictive or diagnostic for VTE development in cancer patients undertaking chemotherapy.
3. Identify differences in these potential biomarkers between cancer patients and volunteers without cancer at baseline assessment.
4. Assess the performance of the Khorana score, Khorana/Ay score (The Vienna model) and PROTECHT score as risk assessment models (RAMs) for predicting VTE development in cancer patients undertaking chemotherapy.
5. Investigate the utility of a calibrated, automated fluorogenic thrombin generation assay in predicting/diagnosing VTE development in cancer patients, at baseline and undertaking chemotherapy.
6. Investigate changes in thrombomodulin and antithrombin levels during chemotherapy and their association with VTE in cancer patients undertaking chemotherapy.
7. Investigate ratios of thrombin generation variables, thrombomodulin and or antithrombin in cancer patients undertaking chemotherapy and their association with VTE.
8. Investigate changes in angiopoietin-1, angiopoietin-2 and/or soluble Tie-2 receptor levels in cancer patients undertaking chemotherapy and their association with VTE.
9. Investigate ratios of angiopoietin-1, angiopoietin-2 and soluble Tie-2 receptor levels with each other in cancer patients undertaking chemotherapy and their association with VTE.
Methods
A prospective, observational study was performed, through the Canterbury Regional Cancer and Haematology Service, to determine the baseline prevalence and cumulative incidence of clinically suspected and unsuspected VTE during chemotherapy treatment. Clinicopathological findings and published risk assessment models (RAMs) were analysed to assess their utility in VTE diagnosis or risk prediction. Adult patients with solid tumour malignancies, lymphoma or myeloma were consented to undergo serial clinical assessments, blood tests and radiological imaging. Exclusion criteria included patients who had received chemotherapy or radiation treatments in the preceding three months or were on anticoagulation except for aspirin or clopidogrel. Clinicopathological findings were compared with age- and gender-matched volunteers without cancer, including plasma levels of variables of a calibrated, automated thrombin generation assay (TGA) and the angiopoietin (Ang)-Tie2 angiogenesis pathway.
Results
Between July 2011 and August 2013, 203 cancer patients and 50 volunteers without cancer were recruited. Fifty one (25.1%) cancer patients were diagnosed with VTE, including 17 (8.4% prevalence) patients at baseline assessment, prior to commencing chemotherapy. All baseline VTEs were clinically unsuspected including 12 PEs with the most proximal site of thrombus being the lobar pulmonary arteries in 5 patients. Thirty one patients undertaking chemotherapy were subsequently diagnosed with VTE during the follow up period. Overall, 60% of the VTE events were clinically unsuspected. RAMs for VTE development showed high specificity and negative predictive value but were not sensitive with a low positive predictive value and were unable to predict clinically unsuspected VTE development. Adaptations of the RAMs, using local population-derived cut-offs, did not significantly improve the precision of the models. The Christchurch percentile adaptation of the Khorana/Ay RAM, however, was the only model in which none of the patients classified as low risk, subsequently developed VTE on chemotherapy.
The D-dimer was a strong marker of VTE risk and development with higher baseline plasma levels seen in patients with cancer compared with volunteers without cancer. Plasma D-dimer concentrations ≥270ng/mL may predict for the development of VTE in cancer patients on chemotherapy within 100 days and, in combination with an Eastern Collaborative Oncology Group performance status >1, may aid in diagnosing VTE prior to commencing chemotherapy. Any increase in D-dimer concentration, after serial testing, was associated with VTE development on chemotherapy with a one-off concentration of ≥355ng/mL associated with VTE diagnosis (sensitivity 100%, specificity 75%). Plasma levels of TGA and the Ang-Tie2 pathway variables were different in cancer patients compared with volunteers without cancer and changed during chemotherapy treatment, but only a decrease in the angiopoietin-2:soluble Tie2 (sTie-2) ratio, after serial testing, was independently associated with VTE development on chemotherapy.
Conclusions
VTE risk is high in cancer patients with many patients presenting with significant clot burden that is clinically unsuspected. Although low risk patients can be reliably identified using RAMs, it is challenging to identify high risk patients in order to tailor thromboprophylaxis. Assessment of the ECOG performance status and plasma concentrations of the D-dimer and Ang-2:sTie2 ratio in cancer patients may be associated with VTE development and warrant further investigation.