The HAS-RISC Score for Venous Thromboembolism (VTE) Risk Stratification in Newly Diagnosed Multiple Myeloma Patients Starting Immunomodulatory Drugs (IMID)

Introduction: Patients with newly diagnosed multiple myeloma (MM) have high risk of venous thromboembolism (VTE) when starting initial treatment that contains immunomodulatory drugs (IMID) such as lenalidomide or thalidomide. The National Comprehensive Cancer Network (NCCN) guideline recommends primary anticoagulant thromboprophylaxis for the high-risk patients. However, it is challenging to risk-stratify patients without a validated risk model. We have conducted a retrospective cohort study using the SEER-Medicare (Surveillance, Epidemiology, and End Results) database to derive a new VTE risk assessment model.

Methods: We selected all patients 66 or older with newly diagnosed MM 2007 to 2013. Patients were included if they had a prescription of IMID within twelve months of diagnosis and complete enrollment for fee-for-service and prescription drug coverage. We ascertained baseline demographics and VTE risk factors from the current NCCN guideline using validated codes. The VTE outcome was defined as either one inpatient or two outpatient claims at least 30 days apart in combination with an anticoagulant prescription within 90 days. All patients were followed from the date of IMID initiation until first VTE occurrence or death and were censored for disenrollment from Medicare, discontinuation of IMID (after a grace period of 90 days), autologous transplantation, or the end of claims data (12/31/2014).

Cause specific Cox regression models were used for time to VTE analysis. For variable selection, all risk factors with p-value <0.10 were considered candidates for inclusion in the final multivariable regression model. VTE history, recent surgery, and anticoagulant exposure were forced into the model, regardless of significance testing. Integer points were assigned according to the beta coefficients and subsequent risk groups were created. The model's discrimination was validated internally by the bias-corrected Harrell's c statistic and the 95% confidence interval was estimated from 200 bootstrap samples.

Results: We identified 2397 MM patients on IMID that met the study criteria. The median time on IMID treatment was 116 days (IQR 28-279). The mean age of patients was 74, 49% were female, 80% were White, 13% were Black, 6.5% were Asian. Only 13% of patients had concurrent anticoagulant exposure (11% warfarin, 2% LMWH, 1% DOAC) with a median duration of 116 days (IQR 42-315 days). In the multivariable model built from candidate covariates, we identified history of VTE, recent surgery, cytotoxic (non-bortezomib) chemotherapy, higher dose dexamethasone, older age, and Black race, as important risk factors. Asian race and LMWH/DOAC use were associated with lower VTE risk (Table 1).

We derived a risk assessment model that stratified patients into 2 prognostic risk groups (Table 1): 25% (n=581) in the very high-risk group (score 2 to 7), 75% (n=1816) in the standard-risk group (score -3 to 1). The incidence of VTE at 3 months and 6 months were 9.5% and 16.3% in the very high-risk group compared to 3.7% and 6.3% in the standard-risk group with a resulting hazard ratio of 2.73 (p<0.001) (Figure 1). The bias-corrected Harrell's c statistic for the product index was 0.63 (0.59-0.68).

Conclusions: We have derived a VTE risk assessment model specifically for patients with MM starting IMID therapy. The HAS-RiSC score combines 7 clinical risk factors - History of VTE, Age 80+, Surgery within last 90 days, Race Black, race Asian, Steroid use, and Chemotherapy - into a simplified VTE risk assessment model that identifies a subgroup of patients at very high risk for VTE. External validation of this risk assessment model is currently in progress.

View largeDownload slide

View largeDownload slide

Close modal