A Novel Score for Prediction of Cancer in Unprovoked Venous Thromboembolism

Introduction

Venous thromboembolism (VTE) can be the first sign of an underlying cancer. Current guidelines recommend limited cancer screening for patients with unprovoked VTE. However, some patients may have a higher risk of cancer, and identifying these individuals could help direct physicians on who would benefit from more extensive screening. This study aims to determine the incidence and predictors of cancer within 12 months following an unprovoked VTE.

Method

We conducted a retrospective analysis of patients diagnosed with unprovoked VTE, defined according to ISTH criteria, from January 2016 to December 2022. VTE includes all pulmonary embolisms, deep vein thrombosis, mesenteric and cerebral thrombosis. Patients with provoked VTE, cancer at diagnosis, and those who were lost to follow-up were excluded. Extensive cancer screening was defined as undergoing any of the following imaging modalities: CT chest, abdomen, pelvis, ultrasound abdomen, PET scan, or MRI. Data were collected through manual chart review.

To identify predictors of cancer diagnosis, we first performed univariate logistic regression analyses to assess the significance of individual predictors. Variables with a p-value < 0.1 in the univariate analysis were then included in a multivariate logistic regression model to evaluate the combined effect of multiple predictors. The variables analyzed included age, gender, BMI, comorbidities (diabetes mellitus, hypertension, stroke, chronic lung disease, smoking status, chronic kidney disease, previous cancer, ischemic heart disease, congestive heart failure, family history of cancer), and initial laboratory values (hemoglobin, platelet count, and white blood cell count). In the multivariate analysis, a significance level of p < 0.05 was used to identify statistically significant predictors. Statistical analyses were conducted using SPSS version 29.0.2. The study was approved by the Institutional Review Board, and informed consent was waived.

Results

After exclusion, a total of 447 patients were included in the study. The median age of the patients was 60 years (IQR 22), with 41 % being male. Current smokers comprised 4.3% of the patients, and past smokers comprised 1 %. One hundred eighty patients (40.3%) underwent extensive screening, and 267 (59.7%) did not undergo extensive screening. Within 12 months, 24 patients (5.37%) were diagnosed with cancer. Among the patients who had extensive screening, 21 (11.67%) were diagnosed with cancer compared to 3 (1.12%) in the non-screened group. Significant differences were observed between the screened and non-screened groups in terms of age and hypertension. The median age was 60 years (IQR 22) in the screening group and 53 years (IQR 27) in the non-screening group (p = 0.005). Hypertension was present in 88 patients (48.8%) in the screening group and 98 patients (36.5%) in the non-screening group (p = 0.013). Regarding the reasons for screening, 258 patients (60.4%) underwent screening without symptoms, and 169 patients (39.6%) were screened due to symptoms. Univariate analysis identified age (p = 0.023) and first hemoglobin level (p = 0.034) as significant predictors of cancer diagnosis. Multivariate analysis confirmed age as a significant predictor (coefficient = 0.0325, p = 0.021), with first hemoglobin level showing borderline significance (coefficient = -0.0088, p = 0.064). Based on our analysis, we developed a logistic regression model to assess the risk of cancer diagnosis within 12 months following a VTE diagnosis. The model uses patient age and hemoglobin levels as predictors.

The logistic regression equation derived from our analysis is: Risk Score=−3.6317+0.0325×Age−0.0088×Hemoglobin Level. The risk score is then converted to a probability using the logistic function. The calculator based on this model is available online at https://mzn.edu.sa/sn/

Conclusion

We developed a novel and simple method for predicting cancer in patients with unprovoked VTE using two variables: age and hemoglobin level. This score is kept continuous rather than dichotomized, as the definition of high versus low risk is arbitrary. External validation of this score is necessary to confirm its predictive ability for diagnosing cancer at 12 months. This model may assist in predicting the risk of cancer and help determine which patients would benefit from extensive screening.

No relevant conflicts of interest to declare.