Soluble P-Selectin for the Diagnosis of Deep Vein Thrombosis (DVT) in Cancer Patients

Accurate and early diagnosis of DVT is crucial to prevent untreated DVT from progressing to fatal PE. Clinical scoring systems used in combination with D-dimer for DVT diagnosis can exclude but not confirm DVT, and vascular ultrasound is still needed. Diagnostic algorithms are challenging in cancer, where the burden of VTE is higher, clinical scores less accurate and baseline D-dimer frequently elevated. Soluble P-selectin (sPsel) is expressed upon activation of platelets and endothelial cells by a thrombotic stimulus. sPsel is known to be important in predicting VTE in cancer patients. We previously showed that elevated sPsel in combination with Wells' score can confirm lower extremity (LE) DVT with a positive predictive value (PPV) of 100% in a general patient cohort (Clin Appl Thromb Hemost, 2011; 17:425–31). To date, no validated biomarker can establish VTE in cancer patients. The primary objective of this analysis was to identify a biomarker that alone or in combination confirms the presence of acute DVT in cancer patients. Secondary objective was to apply the previously derived model to patients with active cancer.

Patients referred to the diagnostic vascular ultrasound lab with signs or symptoms concerning for acute LE DVT were enrolled in a prospective cohort study. Following informed consent, patients underwent LE doppler, clinical assessment, and blood draw for biomarkers sPsel, D-dimer, and CRP. Exclusion criteria were age < 18, isolated calf vein DVT, pregnant, nursing, or on therapeutic anticoagulation. Only subjects with active cancer, defined as actively receiving anti-tumor therapy, or with physical or radiological evidence of malignancy within 6 months of enrollment, were included in this subset analysis of previously published data. Continuous and categorical values were compared with t-test and chi-squared test, respectively. PPV and NPV were estimated from cut points.

Between 2006 to 2012, 442 patients with LE DVT were eligible for analysis; 99 had active cancer. 60 cases (63% male) had a confirmed proximal LE DVT; 39 controls (56% male) had leg pain but negative for VTE. Cases had increased prior history (p=0.001) and family history of VTE (p=0.039). No difference in mean age (62.4 vs. 61.8 years), BMI, or active chemotherapy was seen. Cancer types (%) included: breast (4), lung (10), hematologic (22), melanoma (3), GI (16), GU (16), Gyn (11), brain (4), other (6). sPsel, D-dimer, CRP, and Wells' score were all elevated in cases versus controls at time of DVT diagnosis.

Application of our previously derived model of sPsel + Wells' with cut point of 2 yielded a similar PPV but lower NPV in active cancer versus non-cancer patients. Had imaging not been available, we could have ruled in or out 27.5% (27/98) patients for LE-DVT with this rule. D-dimer + Wells' had a superior NPV, albeit in a low number of patients. Raising the Wells' cut point to 3 (to accommodate a point assigned in the score for active cancer) improved the PPV for both the sPsel and D-dimer models, at the expense of the NPV in the D-dimer model. 38% (37/98) cancer patients could be ruled in or out for LE-DVT without imaging in this sPsel model.

D-dimer is the most commonly used marker to rule out acute DVT. However, due to its low specificity and elevated baseline value in patients with cancer, it cannot be used to confirm clot presence. In patients with active cancer presenting with potential symptoms of LE DVT, sPsel, D-dimer, CRP and Wells' criteria were all elevated in cases versus controls. A combination of sPsel ≥ 90 ng/mL + Wells' ≥ 3 was superior to the D-dimer + Wells score to rule in clot at the time of presentation. D-dimer remained superior to rule out clot in this population. In the future, rapid VTE diagnosis through the use of novel biomarkers such as sPsel may help improve morbidity and mortality.

No relevant conflicts of interest to declare.