In this issue of Blood, Federici et al present compelling evidence that bleeding score (BS) predicts the risk of future bleeding in von Willebrand disease (VWD) patients.1
Diagnosing VWD, particularly in its milder forms, is a well-recognized challenge. There is no single diagnostic test and no international consensus about diagnostic criteria. Bleeding assessment tools (BATs) have helped meet this challenge by standardizing bleeding histories and quantitating bleeding symptoms. Substantial evidence exists that a positive or abnormal bleeding score is sensitive for the diagnosis of VWD. Substantial evidence also exists that a normal BS essentially rules out a diagnosis of VWD. BSs have been used to successfully describe symptom severity in a number of bleeding disorders and facilitate research evaluating phenotype-genotype interactions.2 Until now, however, there has never been a study addressing the value of a BS in predicting the risk of future bleeding, an aligned, but separate issue from the ability of a BS to predict a diagnosis.
Federici et al report the results of a prospective observational cohort study performed in 6 comprehensive bleeding disorder clinics in Italy. In total, they studied 796 (primarily adult) VWD patients. A BS was calculated using the Molecular and Clinical Markers for the Diagnosis and Management of Type 1 VWD (MCMDM-1VWD) Bleeding Questionnaire at the time of study enrollment and then patients were prospectively followed for 1 year to identify those who experienced spontaneous bleeding severe enough to require treatment with either desmopressin and/or von Willebrand factor/factor VIII concentrates. Of the 796 patients enrolled, 75 (∼10%) needed treatment of 232 spontaneous bleeding episodes. As shown in Table 1, multivariable analysis showed that a BS > 10 was the only variable predictive of the risk of future bleeding (adjusted hazard ratio [HR], 7.27; 95% confidence interval [CI], 3.83-13.83). Lower BSs were not as useful.1 Based on this, the authors conclude that the BS could influence treatment decisions by identifying those patients who would benefit from more intensive therapeutic regimens. Indeed, a number of the patients treated on demand in this study were subsequently shifted to a prophylactic regimen following the 1-year observation period. The data support the authors’ conclusion within the context of the population studied and most certainly for other similar patient populations. However, before widespread adoption of this strategy, a number of caveats need to be considered.
Risk of bleeding in the 796 VWD patients according to clinical and laboratory predictors
| . | Crude HRs (95% CI) . | Adjusted HRs (95% CI)* . |
|---|---|---|
| BS | ||
| <5 | 1† | 1† |
| 5-10 | 2.10 (1.10-3.90) | 2.05 (1.07-3.91) |
| >10 | 6.80 (3.80-12.30) | 7.27 (3.83-13.83) |
| VWF:RCo, IU/dL | ||
| >30 | 1† | 1† |
| 10-30 | 1.51 (0.72-3.14) | 1.16 (0.54-2.47) |
| <10 | 3.27 (1.77-6.06) | 1.12 (0.50-2.51) |
| FVIII:C, IU/dL | ||
| >40 | 1† | 1† |
| 20-40 | 2.07 (1.16-3.69) | 1.52 (0.80-2.90) |
| <20 | 4.20 (2.43-7.26) | 2.20 (1.05-4.62) |
| . | Crude HRs (95% CI) . | Adjusted HRs (95% CI)* . |
|---|---|---|
| BS | ||
| <5 | 1† | 1† |
| 5-10 | 2.10 (1.10-3.90) | 2.05 (1.07-3.91) |
| >10 | 6.80 (3.80-12.30) | 7.27 (3.83-13.83) |
| VWF:RCo, IU/dL | ||
| >30 | 1† | 1† |
| 10-30 | 1.51 (0.72-3.14) | 1.16 (0.54-2.47) |
| <10 | 3.27 (1.77-6.06) | 1.12 (0.50-2.51) |
| FVIII:C, IU/dL | ||
| >40 | 1† | 1† |
| 20-40 | 2.07 (1.16-3.69) | 1.52 (0.80-2.90) |
| <20 | 4.20 (2.43-7.26) | 2.20 (1.05-4.62) |
The effect of each predictor on the bleeding risk was adjusted for that of the other ones and for age and sex, in a multivariable Cox proportional hazard model.
Reference group.
See Table 2 in the article by Federici et al beginning on page 4037.
The BAT used in this study was published by Tosetto et al in 2006 and results in a summative BS that primarily hinges on the need for medical attention and treatment of bleeding symptoms.3 Higher bleeding scores result from more intensive treatment; for example, epistaxis requiring nasal packing scores much higher than nosebleeds that never required medical attention. Therefore, the BS is unlikely to be predictive of the risk of future bleeding in patients with limited access to medical care, either because of geography or resources. Additionally, although tempting to do so, caution should be exercised before extrapolating these results to patients with other bleeding disorders without further study. Finally, the results are not applicable to children or to patients with type 2N VWD because these patient groups were not included in sufficient numbers.
The authors include mention of a couple of limitations in the Discussion. The first is that the period of follow up was relatively short, particularly for those patients with type 1 VWD (457/796). Certainly, a longer period of observation would have been useful, especially when considering age-related increases in von Willebrand factor levels and the still unresolved question of whether or not this ameliorates bleeding symptoms. The second is that study took place in 2008 to 2009; therefore, the BAT currently endorsed by International Society on Thrombosis and Hemostasis (ISTH-BAT)4 had not yet been published. The MCMDM-1VWD Bleeding Questionnaire was the best choice for this study at the time, and given that the ISTH-BAT descended directly from it, the results remain meaningful today.
Overall, this study provides critically important knowledge about the utility of BS. Although the results are intuitive, this does not in any way diminish the study impact.
Conflict-of-interest disclosure: The author declares no competing financial interests.
