Validation of the ONKOTEV Risk Prediction Model for Venous Thromboembolism in Outpatients With Cancer

Development, validation, and clinical utility of risk prediction models for cancer-associated venous thromboembolism: A retrospective and prospective cohort study

Objectives

This study aims to develop cancer-associated venous thromboembolism (CA-VTE) risk prediction models using survival machine learning (ML) algorithms.

Methods

This study employed a double-cohort study design (retrospective and prospective). The retrospective cohort (n ​= ​1036) was used as training set (70.0%, n ​= ​725) and internal validation set (30.0%, n ​= ​311); while the prospective cohort (n ​= ​321) was used as external validation set. Seven survival ML algorithms, including COX regression, classification, regression and survival tree, random survival forest, gradient boosting survival machine tree, extreme gradient boosting survival tree, survival support vector analysis, and survival artificial neural network, were applied to train CA-VTE models.

Results

Univariate analysis and LASSO-COX regression both selected five predictors: age, previous VTE history, ICU/CCU, CCI, and D-dimer. The seven survival ML models (C-index: 0.709-0.760; Brier Score: 0.212-0.243) all outperformed Khorana Score (C-index: 0.632; Brier Score: 0.260) in external validation set. Among all models, the COX_DD model (COX regression ​+ ​D-dimer) performed best. However, ML models and Khorana Score predicted CA-VTE risk on ≥ 7 days of hospitalization with an increase in Brier Score ≥ 0.25, showing poor calibration.

Conclusions

In this study, the CA-VTE risk prediction models developed in seven survival ML algorithms outperformed Khorana Score. Combining with D-dimer can improve model performance. Applying the nomogram based on the optimal COX_DD model allows oncology nurse to reassess CA-VTE risk once a week. The prediction models developed using survival ML algorithms in this study may contribute to the dynamic and accurate risk assessment of CA-VTE for cancer survivors.

Venous thromboembolic risk in hematological hospitalized patients: a retrospective study

Although patients with hematological malignancies carry a non-negligible risk of venous thromboembolic events (VTE), no risk assessment models (RAMs) have been developed and validated to predict this risk during their hospital stay. Moreover, the performance of RAMs developed to predict the thromboembolic and hemorrhagic risk in hospitalized medical patients has never been assessed in this specific population. In this observational study we sought to assess the incidence of VTE and hemorrhagic events (HE), to evaluate pharmacological prophylaxis use, and to test existing RAMs for TE and HE (Padua Predictions Score, PPS, and Improve Bleeding Score, IBS, respectively) in a population of 514 hematological patients at their first hospital admission. 165 patients were found to be at high thromboembolic and 114 at high hemorrhagic risk according to PPS and IBS; a total of 148 patients received pharmacological prophylaxis with low-molecular weight heparin (LMWH). An incidence of VTE (3.11%) and HE (1.75%) was found, with most thrombotic (15/16) and hemorrhagic (8/9) events occurring in patients not receiving prophylaxis. At the univariate analysis, the diagnosis of acute leukemia was significantly associated with an increased risk of VTE, whereas prophylaxis was found to be protective. However, at the multivariate analysis the statistical significance was lost for both of these variables (HR = 2.76, p = 0.218 and HR = 0.22, p = 0.170, respectively). PPS poorly performed in this population, but no other factors associated with VTE were found. Larger, multicenter studies to develop better RAMs tailored to HM patients are needed to improve VTE management in this population.

Relevance of recurrent venous thromboembolism according to location of metastasis in patients with cancer-associated thrombosis. A cohort study

BACKGROUND

Risk of VTE recurrence (VTEr) in patients with cancer-associated thrombosis (CAT) is high. Cancer-related risk factors for VTEr have been studied, but information about the importance of location of metastasis is scarce.

OBJECTIVES

1) Evaluate rate of VTEr in CAT patients according to location of metastasis, and 2) Identify variables associated to VTEr during long-term follow-up.

METHODS

A retrospective, single-center, non-interventional study of consecutives patients with CAT conducted between 2007 and 2022. Haematological neoplasms were excluded.

RESULTS

Among 1248 patients with CAT (age 64.1 ± 12.8 years; 48.2 % female) followed-up for 13.19 months (p25-75, 5.6-26.9) there were 141 VTEr. The rate of VTE recurrence in patients without and with metastasis were 4.72 per 100 patient-years (95%CI: 3.66-6) and 10.05 per 100 patient-years (95 % CI: 7.89-12.61), respectively. The metastases locations associated with VTEr, compared to those without metastasis, were lung (rate ratio [RR]: 2.21; 95 % CI: 1.42-3.43), liver (RR: 2.02; 95%CI: 1.26-3.24), pancreas (RR: 6.21; 95 % CI: 1.52-25.35), pleura (RR: 2.93; 95%CI: 1.58-5.41), bone (RR: 2.16; 95 % CI: 1.29-3.64) and adrenal (RR: 6.18; 95%CI: 2.97-12.86). Multivariate analysis of variables associated with VTEr beyond 12 months were male sex (hazard ratio [HR] 1.54, 95%CI: 1.08-2.19), ECOG performance status >1 (HR 1.74, 95%CI: 1.03-2.94), metastasis in 1-2 locations (HR 2.38, 95%CI: 1.68-3.37) and metastasis in >2 locations (HR 3.88, 95%CI: 1.68-8.98).

CONCLUSIONS

The rate of VTEr differs according to the location of metastasis. We identified variables related to VTEr during long-term follow-up which may help clinicians decide whether to continue anticoagulation.

SEOM clinical guidelines on venous thromboembolism (VTE) and cancer (2023)

The Spanish Society of Medical Oncology (SEOM) last published clinical guidelines on venous thromboembolism (VTE) and cancer in 2019, with a partial update in 2020. In this new update to the guidelines, SEOM seeks to incorporate recent evidence, based on a critical review of the literature, to provide practical current recommendations for the prophylactic and therapeutic management of VTE in patients with cancer. Special clinical situations whose management and/or choice of currently recommended therapeutic options (low-molecular-weight heparins [LMWHs] or direct-acting oral anticoagulants [DOACs]) is controversial are included.

Comparison of Khorana vs. ONKOTEV predictive score to individualize anticoagulant prophylaxis in outpatients with cancer

BACKGROUND

Based on the Khorana score, guidelines recommend anticoagulation for primary prophylaxis (PP) in outpatients with cancer with an intermediate-to-high risk of venous thromboembolism (VTE). ONKOTEV score has been prospectively externally validated as novel risk assessment model (RAM) with good discriminatory performances but no direct comparisons with Khorana Score are available.

METHODS

Using the ONKOTEV validation dataset (n = 425), we applied generalized decision curve analysis (gDCA) which integrates the principles of evidence-based medicine with treatment effects, model accuracy and patient preferences (weighted as the relative value [RV] of avoiding VTE versus major bleeding [MB]). The aim is to select the most optimal treatment strategy among multiple options: "no treatment", "treat all patients with DOAC/LMVH", or "use ONKOTEV/KHORANA scores to guide PP with DOAC/LMWH".

RESULTS

Results showed that ONKOTEV-guided PP (using DOAC or LMWH) remained the most optimal strategy for wide range assumption of treatment efficacy and patient's preference. For those patients, who value avoiding VTE more than MB, then offering DOAC to all patients represents the best strategy. When MBs are feared more than the morbidity of VTE, ONKOTEV-guided PP (DOAC) represents the best management strategy. In all cases, ONKOTEV outperformed Khorana for individualized VTE prevention.

CONCLUSIONS

When the two predictive models are integrated within a decision analysis framework, ONKOTEV appears superior to Khorana Score in guiding individualized prevention of cancer-related VTE in outpatients with cancer. The findings herein reported provide cutting edge insights in cancer care and support the spread of ONKOTEV score in the ambulatory cancer setting.

Decision theoretical foundations of clinical practice guidelines: an extension of the ASH thrombophilia guidelines

ABSTRACT

Decision analysis can play an essential role in informing practice guidelines. The American Society of Hematology (ASH) thrombophilia guidelines have made a significant step forward in demonstrating how decision modeling integrated within Grading of Recommendations Assessment, Developing, and Evaluation (GRADE) methodology can advance the field of guideline development. Although the ASH model was transparent and understandable, it does, however, suffer from certain limitations that may have generated potentially wrong recommendations. That is, the panel considered 2 models separately: after 3 to 6 months of index venous thromboembolism (VTE), the panel compared thrombophilia testing (A) vs discontinuing anticoagulants (B) and testing (A) vs recommending indefinite anticoagulation to all patients (C), instead of considering all relevant options simultaneously (A vs B vs C). Our study aimed to avoid what we refer to as the omitted choice bias by integrating 2 ASH models into a single unifying threshold decision model. We analyzed 6 ASH panel's recommendations related to the testing for thrombophilia in settings of "provoked" vs "unprovoked" VTE and low vs high bleeding risk (total 12 recommendations). Our model disagreed with the ASH guideline panels' recommendations in 4 of the 12 recommendations we considered. Considering all 3 options simultaneously, our model provided results that would have produced sounder recommendations for patient care. By revisiting the ASH guidelines methodology, we have not only improved the recommendations for thrombophilia but also provided a method that can be easily applied to other clinical problems and promises to improve the current guidelines' methodology.

Gynecological Cancer and Venous Thromboembolism: A Narrative Review to Increase Awareness and Improve Risk Assessment and Prevention

The prevention and appropriate management of venous thromboembolism in cancer patients is of paramount importance. However, the literature data report an underestimation of this major problem in patients with gynecological cancers, with an inconsistent venous thromboembolism risk assessment and prophylaxis in this patient setting. This narrative review provides a comprehensive overview of the available evidence regarding the management of venous thromboembolism in cancer patients, focusing on the specific context of gynecological tumors, exploring the literature discussing risk factors, risk assessment, and pharmacological prophylaxis. We found that the current understanding and management of venous thromboembolism in gynecological malignancy is largely based on studies on solid cancers in general. Hence, further, larger, and well-designed research in this area is needed.

In Search of the Perfect Thrombosis and Bleeding-Associated Cancer Scale

Thrombosis and bleeding are commonly observed in cancer patients, and their management is crucial for positive patient outcomes. A comprehensive, prophylactic, and therapeutic management of venous thrombosis should focus on identifying the patients who would benefit most from treatment to reduce mortality and minimize the risk of thrombosis recurrence without significantly increasing the risk of bleeding. Existing cancer scales provide valuable information for assessing the overall burden of cancer and guiding treatment decisions, but their ability to predict thrombotic and bleeding events remains limited. With increasing knowledge of the pathophysiology of cancer and the availability of advanced anticancer therapies, new risk factors for cancer-associated thrombosis and bleeding are being identified. In this report, we analyze the current literature and identify new risk factors for venous thrombosis and bleeding which are not included in routinely used risk scores. While some existing cancer scales partially capture the risk of thrombosis and bleeding, there is a need for more specific and accurate scales tailored to these complications. The development of such scales could improve risk stratification, aid in treatment selection, and enhance patient care. Therefore, further research and development of novel cancer scales focused on thrombosis and bleeding are warranted to optimize patient management and outcomes.

Decision curve analysis based on summary data

BACKGROUND

To realize the potential of precision medicine, predictive models should be integrated within the framework of decision analysis, such as the decision curve analysis (DCA). To date, its application has required individual patient data (IPD) that are often unavailable. Performing DCA using aggregate data without requiring IPD may advance the goals of precision medicine.

METHODS

We present a statistical framework demonstrating that DCA can be conducted by using only the mean and standard deviation (SD) from the raw probabilities of the predictive model. We tested our theoretical framework by performing extensive simulations and comparing the aggregate-based DCA with IPD DCA. The latter was conducted using IPD from four predictive models that employed logistic regression, Cox or competing risk time-to-event modeling including (a) statins for primary prevention of cardiovascular disease (n = 4859), (b) hospice referral for terminally ill patients (n = 9104), (c) use of thromboprophylaxis for preventing venous thromboembolism in patients with cancer (n = 425) and (d) prevention of sinusoidal obstruction syndrome after hematopoietic cell transplantation (SCT) (n = 80).

RESULTS

Simulations assuming perfect calibration showed that regardless of which probability distributions informed the predictive models, the differences in DCA were negligible. Similarly, for the adequately powered models, the results of DCA based on the summary data were similar to IPD-derived DCA. The inherent instability of the predictive models, based on the smaller sample sizes, resulted in a somewhat larger discrepancy between aggregate and IPD-based DCA.

CONCLUSIONS

DCA informed by adequately powered and well-calibrated models using only summary statistical estimates (mean and SD) approximates well models using IPD. Use of aggregate data will facilitate broader integration of predictive with decision modeling toward the goals of individualized decision-making.

Novel Insights in Venous Thromboembolism Risk Assessment Methods in Ambulatory Cancer Patients: From the Guidelines to Clinical Practice

Many cancer patients will experience venous thromboembolism (VTE) at some stage, with the highest rate in the initial period following diagnosis. Novel cancer therapies may further enhance the risk. VTE in a cancer setting is associated with poor prognostic, a decreased quality of life, and high healthcare costs. If thromboprophylaxis in hospitalized cancer patients and perioperative settings is widely accepted in clinical practice and supported by the guidelines, it is not the same situation in ambulatory cancer patient settings. The guidelines do not recommend primary thromboprophylaxis, except in high-risk cases. However, nowadays, risk stratification is still challenging, although many tools have been developed. The Khrorana score remains the most used method, but it has many limits. This narrative review aims to present the current relevant knowledge of VTE risk assessment in ambulatory cancer patients, starting from the guideline recommendations and continuing with the specific risk assessment methods and machine learning models approaches. Biomarkers, genetic, and clinical features were tested alone or in groups. Old and new models used in VTE risk assessment are exposed, underlining their clinical utility. Imaging and biomolecular approaches to VTE screening of outpatients with cancer are also presented, which could help clinical decisions.

Ambulatory cancer patients: who should definitely receive antithrombotic prophylaxis and who should never receive

Up to 15-20% of cancer patients experience one or more episodes of venous thromboembolism during cancer disease. Approximately 80% of all cancer-associated venous thromboembolic events occur in non-hospitalized patients. Routine thromboprophylaxis for outpatients with cancer who start new anticancer treatment is currently not recommended by the international guidelines due to the high heterogeneity of these patients in terms of VTE or bleeding risks, the difficulties in selecting patients at high risk, and the uncertainty of duration of prophylaxis. Although the international guidelines endorsed the Khorana score for estimating the thrombotic risk in ambulatory cancer patients, the discriminatory performance of this score is not completely convincing and varies according to the cancer type. Consequently, a minority of ambulatory patients with cancer receive an accurate screening for primary prophylaxis of VTE. The aim of this review is to provide support to physicians in identifying those ambulatory patients with cancer for whom thromboprophylaxis should be prescribed and those that should not be candidate to thromboprophylaxis. In absence of high bleeding risk, primary thromboprophylaxis should be recommended in patients with pancreatic cancer and, probably, in patients with lung cancer harboring ALK/ROS1 translocations. Patients with upper gastrointestinal cancers are at high risk of VTE, but a careful assessment of bleeding risk should be made before deciding on antithrombotic prophylaxis. Primary prevention of VTE is not recommended in cancer patients at increased risk of bleeding as patients with brain cancer, with moderate-to-severe thrombocytopenia or severe renal impairment.

Prediction of Venous Thromboembolism in Patients With Cancer Using Machine Learning Approaches: A Systematic Review and Meta-Analysis

PURPOSE

Recent studies have suggested that machine learning (ML) could be used to predict venous thromboembolism (VTE) in cancer patients with high accuracy.

METHODS

We aimed to evaluate the performance of ML in predicting VTE events in patients with cancer. PubMed, Web of Science, and EMBASE to identify studies were searched.

RESULTS

Seven studies involving 12,249 patients with cancer were included. The combined results of the different ML models demonstrated good accuracy in the prediction of VTE. In the training set, the global pooled sensitivity was 0.87, the global pooled specificity was 0.87, and the AUC was 0.91, and in the test set 0.65, 0.84, and 0.80, respectively.

CONCLUSION

The prediction ML models showed good performance to predict VTE. External validation to determine the result's reproducibility is necessary.