Predicting major bleeding during extended anticoagulation for unprovoked or weakly provoked venous thromboembolism

Predicting intracranial bleeding during anticoagulation for venous thromboembolism within different time frames: Findings from the RIETE registry

BACKGROUND

The risk of intracranial bleeding during anticoagulation for venous thromboembolism (VTE) is substantial and persists beyond the initial treatment phase. We aimed to refine risk-assessment through phase-specific prognostic scores.

METHODS

We identified data from 77,786 VTE patients in the RIETE registry from March 2009 to October 2023 to develop two prognostic scores for intracranial bleeding. Multivariable Cox regression was used to analyze distinct variables for the early (≤90 days) and late (>90 days) phases, with comparative validation against existing scores (modified ACCP, RIETE, VTE-BLEED, and CHAP).

RESULTS

Intracranial bleeding occurred in 411 patients (0.53 %), with 208 cases in the early phase and 203 in the late phase. The 30-day mortality was 45 % and 35 %, respectively. Shared significant predictors for both phases include baseline abnormal mental status, brain cancer, recent intracranial bleeding, and epilepsy. Unique to early-phase bleeding were body weight, non-brain cancer, hypertension, dementia, thrombocytopenia, renal insufficiency, and thrombolytic therapy. Advanced age, pulmonary embolism initially, prior stroke, depression, treatment with direct oral anticoagulants, and use of corticosteroids predicted late-phase bleeding. Both prognostic scores showed a c-statistic of 0.68, outperforming existing scores.

CONCLUSIONS

The study introduces two temporal prognostic scores for intracranial bleeding during anticoagulation for VTE. By discerning specific risk factors pertinent to each treatment phase, these scores outperform traditional models, offering an advanced tool for clinical decision-making. They hold significant potential for optimizing anticoagulation management and reducing bleeding-related mortality.

Baseline platelet count and long-term clinical outcomes in patients with acute venous thromboembolism: a prospective cohort study

An abnormal platelet count (PC) is common in acute venous thromboembolism (VTE) but its relationship with clinical outcomes remains ill-defined. We aimed to explore the association between baseline PC and the long-term risk of clinically relevant outcomes in a prospective cohort of 991 patients with acute VTE. We classified patients into four PC groups: very low (< 100 G/l), low (≥ 100 to < 150 G/l), normal (≥ 150 G/l to ≤ 450 G/l), and high (> 450 G/l). The primary outcome was major bleeding (MB), secondary outcomes were recurrent VTE and overall mortality. We examined the association between PC and clinical outcomes, adjusting for confounders, competing risk for mortality, and periods of anticoagulation. After a median follow-up of 30 months, 132 (13%) of patients experienced MB, 122 (12%) had recurrent VTE, and 206 (21%) died. Compared to patients with a normal PC, patients with a very low PC had a sub-distribution hazard ratio (SHR) for MB of 1.23 (95% confidence interval [CI] 0.52-2.91) and those with a high PC a SHR of 1.87 (95%CI 0.82-4.29). Patients with a low PC had a twofold increased VTE recurrence risk (SHR 2.05, 95%CI 1.28-3.28). Patients with low and very low PC had a hazard ratio for mortality of 1.43 (95%CI 0.99-2.08) and of 1.55 (95%CI 0.80-2.99), respectively. Our findings do not suggest a consistent relationship between baseline PC and long-term clinical outcomes in patients with VTE.

Risk assessment tools for bleeding in patients with unprovoked venous thromboembolism: an analysis of the PLATO-VTE study

BACKGROUND

Guidelines suggest indefinite anticoagulation after unprovoked venous thromboembolism (VTE) unless the bleeding risk is high, yet there is no consistent guidance on assessing bleeding risk.

OBJECTIVES

This study aimed to evaluate the performance of 5 bleeding risk tools (RIETE, VTE-BLEED, CHAP, VTE-PREDICT, and ABC-Bleeding).

METHODS

PLATO-VTE, a prospective cohort study, included patients aged ≥40 years with a first unprovoked VTE. Risk estimates were calculated at VTE diagnosis and after 3 months of treatment. Primary outcome was clinically relevant bleeding, as per International Society on Thrombosis and Haemostasis criteria, during 24-month follow-up. Discrimination was assessed by the area under the receiver operating characteristic curve (AUROC). Patients were classified as having a "high risk" and "non-high risk" of bleeding according to predefined thresholds; bleeding risk in both groups was compared by hazard ratios (HRs).

RESULTS

Of 514 patients, 38 (7.4%) had an on-treatment bleeding. AUROCs were 0.58 (95% CI, 0.48-0.68) for ABC-Bleeding, 0.56 (95% CI, 0.46-0.66) for RIETE, 0.53 (95% CI, 0.43-0.64) for CHAP, 0.50 (95% CI, 0.41-0.59) for VTE-BLEED, and 0.50 (95% CI, 0.40-0.60) for VTE-PREDICT. The proportion of high-risk patients ranged from 1.4% with RIETE to 36.9% with VTE-BLEED. The bleeding incidence in the high-risk groups ranged from 0% with RIETE to 13.0% with ABC-Bleeding, and in the non-high-risk groups, it varied from 7.7% with ABC-Bleeding to 9.6% with RIETE. HRs ranged from 0.93 (95% CI, 0.46-1.9) for VTE-BLEED to 1.67 (95% CI, 0.86-3.2) for ABC-Bleeding. Recalibration at 3-month follow-up did not alter the results.

CONCLUSION

In this cohort, discrimination of currently available bleeding risk tools was poor. These data do not support their use in patients with unprovoked VTE.

Poor patients' knowledge about venous thromboembolism and its therapy is associated with increased risk of major bleeding and discontinuation of anticoagulation: A cohort study

It has been shown that patients' knowledge about venous thromboembolism (VTE) and its therapy is suboptimal, which might reduce compliance and worsen prognosis. We investigated whether low VTE patients' knowledge affects their clinical outcomes during long-term follow-up. We evaluated 151 consecutive patients (51.8 ± 15.7 years) after unprovoked VTE, who were recruited from the outpatient clinic (Krakow, Poland). All patients received anticoagulant treatment, mostly with direct oral anticoagulants (n = 113, 74.8%). The modified Jessa Atrial fibrillation Knowledge Questionnaire (JAKQ-VTE; 16 questions) was used to assess the knowledge of VTE and anticoagulant therapy. During a median follow-up of 58.0 months, VTE recurrence, major bleeding, and anticoagulation withdrawal were recorded. The median percentage of correct responses was 62.5% (12.5-100%) and was inversely correlated with age (P < .01). Diabetic patients and those with positive family history of VTE had lower overall scoring compared to the remainder (both P < .05). Major bleeding (n = 10, 6.6%) and anticoagulation withdrawal (n = 28, 18.5%), but not VTE recurrence (n = 12, 7.9%), were associated with lower overall scoring compared to the remainder (48.8% ± 12.5% vs 63.8% ± 16.3%, P = .003 and 55.3% ± 14.7% vs 64.4% ± 16.3%, P = .040, respectively). Major bleeding was independently associated with the female sex (hazard ratio [HR] 6.18; 95% confidence interval [CI] 1.15-33.19, P = .034), younger age (HR per 10 years 0.55; 95% CI 0.34-0.90, P = .016), OAC therapy discontinuation (HR 6.69; 95% CI 1.62-27.70), and lower overall scoring of JAKQ-VTE (HR 0.60 per 10 percentage points; 95% CI 0.40-0.92, P = .019). Insufficient knowledge about VTE and anticoagulant treatment predisposes to a higher risk of major bleeding and therapy discontinuation, but not VTE recurrence in unprovoked VTE patients during long-term follow-up.

A deep-learning approach to predict bleeding risk over time in patients on extended anticoagulation therapy

BACKGROUND

Thus far, all the clinical models developed to predict major bleeding in patients on extended anticoagulation therapy use the baseline predictors to stratify patients into different risk groups. Therefore, these models do not account for the clinical changes and events that occur after the baseline visit, which can modify risk of bleeding. However, it is difficult to develop predictive models from the routine follow-up clinical interviews, which are irregular sequences of multivariate time series data.

OBJECTIVES

To demonstrate that deep learning can incorporate patient time series follow-up data to improve prediction of major bleeding.

METHODS

We used the baseline and follow-up data that were collected over 8 years in a longitudinal cohort study of 2542 patients, of whom 118 had major bleeding. Four supervised neural network-based machine-learning models were trained on the baseline, follow-up, or both datasets using 70% of the data. The performance of these models was evaluated, along with modified versions of 6 previously developed clinical models, on the remaining 30% of the data.

RESULTS

An ensemble of feedforward and recurrent neural networks that used the baseline and follow-up data was the best-performing model, achieving a sensitivity and a specificity of 61% and 82%, respectively, in identifying major bleeding, and it outperformed the previously developed clinical models in terms of area under the receiver operating characteristic curve (82%) and area under the precision-recall curve (14%).

CONCLUSION

Time series follow-up data can improve major bleeding prediction in patients on extended anticoagulation therapy.

Performance of the Simplified Pulmonary Embolism Severity Index in predicting 30-day mortality after acute pulmonary embolism: Validation from a large-scale cohort

BACKGROUND

The performance of existing prognostic scores including the simplified Pulmonary Embolism Severity Index (sPESI) for short-term mortality of non-high-risk PE in Chinese population has not been widely validated.

METHODS

Non-high-risk patients were included from the prospective cohort of the China pUlmonary Thromboembolism REgistry Study (CURES). The sPESI, RIETE, Geneva, modified FAST, and Bova score were validated. The discriminatory performance was measured by the area under the curve (AUC). We also compared the sensitivity, odds ratio, specificity, positive predictive value and negative predictive value of these scores.

RESULTS

A total of 6,873 non-high-risk patients with acute PE were included and 241 (3.5 %) patients died within 30 days. Compared to the Geneva, modified FAST, and Bova score, the AUCs for predicting 30-day death of sPESI and RIETE score were higher at 0.712 (95 % CI, 0.680, 0.743) and 0.723 (95 % CI, 0.691, 0.755) respectively. The sPESI demonstrated the highest sensitivity at 0.809, while the RIETE score, Geneva, Modified FAST and BOVA score showed sensitivities of 0.622, 0.568, 0.477 and 0.502 respectively. A sPESI ⩾1 point was associated with a 4.7-fold increased risk of 30-day all-cause mortality (95 % CI, 3.427, 6.563, p < 0.001), while a RIETE score of ⩾1 point was associated with a 4.5-fold increased risk (95 % CI, 3.127, 6.341, p < 0.001). The Geneva score, modified FAST and Bova score showed inferior performance.

CONCLUSIONS

The implementation of the fewer-parameter, easier-to-calculate sPESI in Chinese patients with PE can help to discriminate patients with extremely low risk of short-term mortality for home treatment or early discharge.

Machine learning analysis of bleeding status in venous thromboembolism patients

Background

Anticoagulation therapy is the mainstay of therapy for patients with venous thromboembolism (VTE). However, continuing or stopping anticoagulants after the first 3 to 6 months is a difficult decision that requires ascertainment of the risk of bleeding and recurrent VTE. Despite the development of several statistical models to predict bleeding, the benefit of machine learning (ML) models has not been investigated in depth.

Objectives

To assess the benefits of ML algorithms in bleeding risk evaluation in VTE patients and gain insight into their baseline information.

Methods

The baseline clinical, demographic, and genotype information was collected for 2542 patients with VTE who were on extended anticoagulation therapy. Six unsupervised dimensionality reduction and clustering ML algorithms were used to visualize and cluster the data for patients with major bleeding (118 patients) and nonbleeders. Eight supervised ML algorithms were trained and compared with the previously derived clinical models using a 5-fold nested cross-validation scheme.

Results

The baseline dataset for bleeders and nonbleeders showed a high degree of similarity. Two novel clusters were discovered within the dataset for bleeders based on the presence of isolated pulmonary embolism or isolated deep vein thrombosis, though the difference in bleeding risks was not statistically significant (P = .32). The supervised analysis showed that the ML and clinical models have similar discrimination (c-statistics, ∼62%) and calibration performance (Brier score, ∼0.045).

Conclusion

The clinical variables recorded at baseline are not distinctive enough to improve bleeding prediction beyond the performance of the existing models, and other strategies or data modalities should be considered.

Balancing risks of recurrent venous thromboembolism and bleeding with extended anticoagulation: a decision analysis

Background

A decision to stop or continue anticoagulation after 3 months of anticoagulation for venous thromboembolism (VTE) should be made by weighing individual risks of recurrence and bleeding.

Objectives

To determine the optimal ratio of recurrence risk reduction to increase the risk of bleeding in terms of maximizing quality-adjusted life years (QALYs) gained.

Methods

Using a microsimulation model, outcomes within 5 years were simulated after assigning extended treatment if absolute recurrence risk reduction outweighed absolute increase in clinically relevant bleeding risk (International Society on Thrombosis and Haemostasis definition), weighted by a certain ratio. Data were simulated based on the Bleeding Risk Study, a prospective cohort including patients after ≥3 months of anticoagulation for unprovoked VTE or provoked VTE with history of VTE. The VTE-PREDICT risk score was used to estimate 5-year risks of recurrent VTE and clinically relevant bleeding.

Results

Among 10,000 individuals (mean age, 60.2 years, 36% female), the ratio of 0.90 (95% CI, 0.51-3.40; ie, bleeding is considered 0.90 the severity of recurrent VTE), with 99% of patients assigned extended anticoagulation, was considered optimal and resulted in 93 (95% CI, -23 to 203) additional QALYs compared with the least favorable ratio (5.10, 0% extended anticoagulation). At the optimal ratio, treatment based on VTE-PREDICT yielded 44 (95% CI, -69 to 157) additional QALYs versus standard of care.

Conclusion

With the current evidence, the optimal ratio between relevant bleeding risk and absolute recurrence risk reduction remains uncertain. Our results confirm that clinical equipoise exists regarding the decision to stop or continue anticoagulation after initial VTE treatment, emphasizing the importance of shared decision-making.

Risk Scores in Venous Thromboembolism Guidelines of ESC, ACCP, and ASH: An Updated Review

Venous thromboembolism (VTE) is associated with high morbidity and mortality. Risk scores associated with VTE have been widely used in clinical practice. Among numerous scores published, those included in guidelines are usually typical risk scores which have been extensively validated and globally recognized. This review provides an updated overview of the risk scores associated with VTE endorsed by 3 guidelines which are highly recognized in the field of VTE including the European Society of Cardiology, American College of Chest Physicians, and American Society of Hematology, focusing on the development, modification, validation, and comparison of these scores, to provide a comprehensive and updated understanding of all the classic risk scores associated with VTE to medical readers including but not limited to cardiologists, pulmonologists, hematologists, intensivists, physicians, surgeons, and researchers. Although each score recommended by these guidelines was more or less validated, there may still be room for further improvement. It may still be necessary to seek simpler, more practical, and more universally applicable VTE-related risk scores in the future.