Risk of Venous Thromboembolism in Hematological Malignancies: The Scandinavian Thrombosis and Cancer Cohort

Background: Several studies have shown a high incidence of venous thromboembolism (VTE) in hematological cancers, comparable with solid cancers, although bleeding is also a prominent complication of the hematological patients. Cancer patients, who develop VTE, have a reduced survival and impaired quality of life if compared to those who do not develop VTE. Hematological cancers are rather rare diseases and most studies have described only some of the entities. Here we want to compare the incidence of VTE in seven subtypes in a large cohort.

Aim: To investigate the risk of VTE in hematological malignancies compared to matched controls in a prospective population based cohort study, the Scandinavian Thrombosis and Cancer (STAC) Cohort.

Methods: TheSTAC Cohort includes 144.952 participants from three population based prospective cohort studies, i.e. The Tromsø Study and the HUNT2 study from Norway, and the Danish Diet, Cancer and Health Study. The participants were enrolled during 1993-1997, and mean follow-up time was 11.7 years. The cohort profile and outcome of first time objectively confirmed VTE events have been described in prior studies. For this study we collected data from the national cancer registries using morphology codes to identify cohort subjects with hematological cancers, divided into 7 groups: multiple myeloma (MM), chronic lymphocytic leukemia (CLL), acute leukemia (myeloid and lymphoblastic) (AL), chronic myeloproliferative neoplasms and myelodysplastic syndrome (CMN/MDS), aggressive non-Hodgkin lymphoma (aggr. NHL), Hodgkin lymphoma (HL), and indolent lymphoma (Ind. L). Subjects with a VTE event more than one year before cancer diagnosis were excluded. For each of the cases 5 controls matched on country, sex and age were identified. We used Cox regression models to estimate the relative risk of VTE across the seven different subtypes of hematological malignancies with a time axis starting one year before the diagnosis of cancer (and similar for matched controls) and ending at a VTE event or end of follow-up. Data were adjusted for age by spline regression.

Results: During follow-up 891 participants were diagnosed with a hematological malignancy, and in this group 41 VTE events were observed corresponding to an incidence of 12.0 events per 1000 person-years (10^-3 p-y). In the control group of 4455 participants 55 VTE events were observed which gave an incidence of events on 2.3* 10^-3 p-y. Having a hematological cancer including all seven investigated types was associated with a six-fold increased risk of developing VTE compared to the matched controls. During follow-up 203 participants were diagnosed with MM, and 10 VTE events were observed giving an event rate of 14.6*10^-3 p-y; hazard ratio (HR) for VTE was 7.2, 95% confidence interval (CI): 3.6-14.3. CLL was diagnosed in 176 cases, and 11 VTE events were observed in this group (event rate 11.5*10^-3 p-y; HR 5.3; 95% CI: 2.7-10.1). Among the 63 participants who were diagnosed with AL during follow-up 2 VTE events were observed corresponding to an event-rate on 12.8*10^-3 p-y; (HR 6.9; 95% CI: 1.7-29.0). In the group of CMN/MDS 4 VTE events were observed among 104 patients (event-rate 12.0*10^-3 p-y; HR 6.4, 95% CI: 2.3-18.0). In aggressive NHL 10 VTE events were observed among 158 patients resulting in an event-rate of 18.9*10^-3 p-y (HR 10.4; 95% CI: 5.2-20.8). Forty-four participants were diagnosed with HL, and 2 VTE events were observed which corresponds to an event-rate of 10.6*10^-3 p-y among these patients (HR 5.1; 95% CI: 1.2-21.4). Indolent lymphoma was diagnosed in 143 subjects, and 2 VTE events were observed (event-rate 3.5*10^-3 p-y; HR 1.9; 95% CI: 0.47-8.0). The results are summarized in the table:

Conclusion: Indolent lymphoma was the only investigated hematological malignancy that was not associated with a significant increased risk of VTE. The other types of hematological malignancies had an increased risk of VTE ranging from approximately 5-10 times, highest in aggressive non-Hodgkin lymphoma and lowest in Hodgkin lymphoma. However a limitation of the study is the small numbers in some of the groups in spite of the large cohort.