Abstract
Cancer is associated with a substantially elevated risk of venous thromboembolism (VTE), but this risk varies considerably between individuals depending on several clinical and laboratory variables. Biomarkers predictive of VTE are needed to better stratify risk. We had previously identified candidate biomarkers predictive of cancer-associated VTE in a substudy of CASSINI, a randomized trial of thromboprophylaxis in people with cancer with Khorana score (KS) of > 2. In this study, we evaluated these biomarkers for prediction of VTE or death in plasma specimens derived from patients enrolled in AVERT, a similarly designed trial of thromboprophylaxis.
We used available baseline plasma samples from 498 of 574 individuals enrolled in AVERT, a prospective randomized trial of apixaban prophylaxis in cancer with KS > 2. Levels of 9 proteins - C-reactive protein (CRP), chemokine ligand 13 (CCL13/MCP-4), N-terminal prohormone of brain natriuretic peptide (NT-proBNP), human growth hormone (hGH), high sensitivity troponin T (hsTnT), C X C motif chemokine 12 (CXCL12/SDF-1), interleukin 1 receptor (IL1R1), thyroid stimulating hormone (TSH), and growth differentiation factor 15 (GDF-15) – were assessed for association with VTE, death, and VTE or death outcomes. Protein concentrations were log₂ transformed and analyzed using cause specific multivariate Cox proportional hazards models adjusted for age, sex, study arm, and Khorana VTE risk score with baseline hazards dependent on cancer type.
The study population comprised 498 patients, evenly split between apixaban and placebo arms. Median age at enrollment was 68 years; 58% were female and 93% were White. The most common cancers were gynecologic (26.7%), lymphoma (25.7%), pancreas (12.4%), lung (9.8%), and gastric (7.8%). Baseline KS was 2 for the majority of patients (N = 325, 65.3%). Of the full cohort, 41 (8.2%) experienced VTE and 60 (12%) died.
In multivariable analysis for VTE, only hsTnT was significantly associated with VTE (p = 0.011); TSH was non-significantly associated (p = 0.09). Patients who developed VTE had higher median hsTnT (11.2 pg/mL) and TSH (1.67 mIU/L) compared to those without VTE (6.9 pg/mL and 1.25 mIU/L, respectively).
In multivariate analysis for mortality, elevated CRP (p = 1.1×10⁻⁵), GDF-15 (p = 0.00071), CCL13 (p = 0.0079), and NT-proBNP (p = 0.0099) were strongly associated with increased mortality. Median levels among those who died versus those who survived were: CRP (16.7 vs. 6.2 mg/L), GDF-15 (2645 vs. 1576 pg/mL), CCL13 (28.4 vs. 21.5 pg/mL), and NT-proBNP (251 vs. 102 pg/mL).
For the composite outcome of VTE or death, CRP (p = 0.000011), GDF-15 (p = 0.0014), and hGH (p = 0.038) were the most significantly associated. Patients who experienced VTE or death had higher median levels compared to those without events: CRP (16.1 vs. 5.8 mg/L), GDF-15 (2413 vs. 1435 pg/mL), NT-proBNP (228 vs. 93 pg/mL), CCL13 (27.4 vs. 20.5 pg/mL), and hGH (0.51 vs. 0.38 ng/mL).Conclusions: Of previously identified candidate biomarkers, only hsTnT was significantly associated with VTE in this analysis of already-high-risk individuals (minimum KS of 2), with higher median levels observed in those who developed VTE. Elevated levels of CRP, GDF-15, NT-proBNP, and CCL13/MCP-4 were significantly associated with increased risk of mortality and the composite outcome of VTE or death, with affected patients showing notably higher median concentrations. These biomarkers need further study for their potential role in identifying individuals at risk for worse outcomes, and to inform personalized strategies for thromboprophylaxis and prognosis in oncology.
