番茄社区

This article is a collaboration between 番茄社区 and:

A simple clinical prediction model has been developed and validated that should help improve physicians' ability to predict which ambulatory patients with solid cancers are most at risk for venous thromboembolism (VTE) and who therefore might most benefit from thromboprophylaxis.

Approximately three-quarters of all cancer-related VTE events occur in ambulatory patients with cancer.

For the research, using data from the prospective Vienna Cancer and Thrombosis Study (CATS), Ingrid Pabinger, MD, of Medical University of Vienna, and colleagues first developed the model based on only one clinical variable -- tumor-site risk category -- and one biomarker -- D-dimer concentrations. The team then validated the model in the prospective Multinational Cohort Study to Identify Cancer Patients at High Risk of Venous Thromboembolism (MICA) to arrive at a nomogram that was able to discriminate between patients who did and did not develop VTE during 6 months of follow-up.

"Decision-curve analysis showed that use of the model to select those patients who would benefit from thromboprophylaxis would provide greater clinical utility by reducing the risks of venous thromboembolism and bleeding events caused by unnecessary thromboprophylaxis, as compared with treat-all or treat-none approaches," the researchers wrote in .

"This novel and simple tool might enable clinicians to identify those ambulatory patients with solid cancers who have a 6-month VTE risk of 10%-15% or more and thus might benefit from thromboprophylaxis, and those with a very low risk of VTE, in whom the increased risk of bleeding due to thromboprophylaxis would outweigh the benefits."

To develop the model, the investigators analyzed data from 1,423 patients in CATS, all of whom had either a solid cancer or a lymphoma.

"The primary outcome of CATS was symptomatic, objectively confirmed, and independently assessed VTE, defined as a composite of distal or proximal deep vein thrombosis (DVT) of the leg, upper limb deep vein thrombosis, symptomatic splanchnic deep vein thrombosis, or pulmonary embolism (PE), occurring during a 2-year observation period," the team reported.

To validate the clinical model, the researchers used demographic, laboratory, and outcome data from 832 patients enrolled in MICA. Patients in that study were followed for a maximum of 6 months or until the occurrence of the same VTE events as evaluated in CATS or the patients died. The median follow-up in CAT was 180 days, during which 6% of patients developed a VTE.

The median follow-up in MICA was similar, ranging from 109 to 180 days, during which an identical proportion of patients, 6%, also developed a VTE.

The cumulative 6-month risk of VTE was 5.7% (95% CI 4.5 to 6.9) in CATS and 6.3% (4.7 to 8.2) in MICA. The most common VTEs in both studies were lower limb DVT and PE. The cross-validated c-index of this model in CATS was 0.66 (95% CI 0.63 to 0.67) and 0.68 (95% CI 0.62 to 0.74) in MICA.

The cutoff for the predicted cumulative 6-month risk of VTE in CATS was set at 10% -- at which time the sensitivity of the model was 33% (95% CI 23 to 47); the specificity was 84% (95% CI 83 to 87); the positive predictive value was 12% (95% CI 8 to 16); and the negative predictive value was 95% (95% CI 94 to 96).

Again at a 10% cutoff for predicted risk of VTE at 6 months in MICA, the sensitivity of the model was 21% (95% CI 10 to 35); the specificity was 87% (95% CI 85 to 90); the positive predictive value was 9% (95% CI 4 to 16); and the negative predictive value was 95% (95% CI 93 to 96).

"Reasonable evidence exists to suggest that thromboprophylaxis halves the absolute risk of venous thromboembolism in patients with cancer," the study authors observed.

Using this estimate, the team calculated that the number needed to treat would be at least 40 for patients with a VTE risk of 2%-5%; 20 to 40 for patients in the 5%-10% VTE risk range; 14 to 19 for patients in the 10%-15% risk range, and fewer than 14 for those whose VTE risk was 15% or higher.

"We posit that thromboprophylaxis is justified for patients with cancer who have a predicted 6-month risk of developing [VTE] of 15% or higher, and perhaps even for those with a 10%-15% risk in light of American College of Clinical Pharmacy guidelines that recommend long-term anticoagulation in patients with VTE in the absence of cancer who have a risk of recurrence of about 10% at 12 months," the investigators suggested.

Study limitations, they said, include the fact that patients in CATS and MICA were recruited from academic centers and as such, may not be representative of all patients with cancer.

Writing in an titled "Simplicity versus Complexity: An Existential Dilemma as Risk Tools Evolve," Alok Khorana, MD, of the Cleveland Clinic Taussig Cancer Institute, called the study important because it aimed to simplify risk assessment for VTE in cancer patients and includes a biomarker, D-dimer, that can easily be tested for in clinics.

Ten years ago, his team published the first formal risk-assessment tool developed to predict the risk of VTE in cancer patients. Now known as the Khorana score, it is the most widely used clinical prediction model aiming to identify ambulatory patients with cancer at increased risk of VTE during chemotherapy by use of tumor site and body-mass index and laboratory measurements of platelets, hemoglobin, and leukocytes.

In the editorial, Khorana pointed out that the new simplified model is applied as a nomogram, and "as such requires clinicians to estimate the risk of [VTE] on a point scale ranging from 0 to 100 ... with further stratification by tumor risk-site, eventually leading to a total points scale of 0 to 200."

These calculations then allow clinicians to arrive at a cumulative 6-month incidence of VTE that ranges from 2% to 29%. "I am concerned about the number of steps that this process asks of busy clinicians," he said, adding, though, that the fact that this new prediction tool can work in an extended setting may make the it helpful in VTE risk assessment of patients who are already receiving chemotherapy.

"My own bias is that applicability is key, and risk stratification should be as automated as possible for optimal clinical use," he concluded.