Diagnostic pitfalls and conundrums in type 1 von Willebrand disease

The diagnosis of von Willebrand disease (VWD) is often challenging and complex due to the variability of plasma von Willebrand factor (VWF) levels in the general population and the difficulties in assessing bleeding phenotype. Type 1 VWD is the most common subtype, with most patients with plasma VWF levels in the 30- to 50-IU/dL range (low VWF). Focusing on type 1 VWD, we discuss some of the common clinical scenarios faced by health care providers and our approach to their management.

A 14-year-old girl was referred from her primary care physician with heavy menstrual bleeding (HMB), mild bruising, and nosebleeds (>10 minutes). She had no relevant family history and no challenges to date. HMB has improved on a low-dose combined oral contraceptive pill (COCP). Hemostatic testing was normal, including plasma VWF:Ag 58 IU/dL and VWF:RCo 53 IU/dL with hemoglobin 10.5 g/dL and ferritin 15 µg/L. Is further testing warranted?

Mild bleeding symptoms are common in the general population, and differentiation from a bleeding disorder is often challenging. Over- and underreporting of bleeding can relate to culture, personality, gender, and educational level. Standardized bleeding assessment tools (BATs) assist in objectively quantifying bleeding symptoms, with the International Society on Thrombosis and Hemostasis (ISTH)–BAT (or self-completed SELF-BAT) commonly used.¹ 

The recent American Society of Hematology/ISTH/World Federation of Haemophilia/National Haemophilia Foundation guidelines relate the use of BATs to the pretest probability of a diagnosis of VWD.²  In the primary care setting (low pretest probability), the high negative predictive value of the bleeding score (BS) identifies those who do not warrant hemostatic testing.²  In the hematology clinic, those referred have either an intermediate (personal bleeding history) or high (affected family member [AFM]) pretest probability of VWD, so the role of the BAT alters, serving now to record bleeding rather than determine testing.²  The value of BATs should not be underestimated; recording BS not only aids standardization and communication between health care providers but also assists in evaluation of bleeding during follow-up (interim bleeding).³  The number of positive domains has also been shown to influence likelihood of VWD diagnosis.^4,5  Use of BATs remains inconsistent, however, with BS completed in only 30% of patients in 1 pediatric study.⁵ 

BATs have specific limitations, saturating easily with the same score for bleeding that is recurrent or a single episode (Figure 1).^1,6  In those with limited hemostatic challenges, lower BS may not represent their phenotype.¹  To address this, a lower threshold is applied for children (ISTH-BAT ≥3 for those <18 years old) than adults (≥4 for males, ≥6 females). This escalation in threshold may impose a barrier to referral for older teenagers with limited challenges. Conversely, in children with multiple but minor bleeding complaints, a positive BS may result despite a mild phenotype, leading to proposals that a higher threshold (ISTH-BAT ≥4) may be more suitable for adolescent females.⁷ 

In our case, based on her personal bleeding history, the ISTH-BAT is 5, meriting hemostatic testing focusing on clotting factor deficiencies, VWD, and qualitative platelet disorders.

Hemostatic testing should include VWF:Ag, platelet-dependent VWF activity assay (VWF:RCo, VWF:GPIbM, or VWF:GPIbR), and FVIII:C. Due to the higher coefficient of variation with VWF:RCo testing, the guidelines suggest the use of newer assays (VWF: GPIbM or VWF:GPIbR) if available.²  They not only correlate well with VWF:RCo but also are insensitive to ristocetin-binding polymorphisms (eg, p.D1472H), which falsely lower VWF:RCo levels but are not associated with bleeding tendency, resulting in spurious VWD diagnoses.⁸ 

VWF is an acute phase reactant, secreted in response to acute or chronic endothelial cell activation or damage (Figure 2) (eg, vascular disease or diabetes).⁹  Epinephrine, elevated in times of physiologic stress, induces VWF release, and transiently elevated plasma VWF levels are seen with exercise, trauma, sepsis, or stress (eg, failed phlebotomy attempts).⁹  Several genetic loci influence plasma VWF levels, which, although outside the scope of this article, have been succinctly reviewed previously.¹⁰  Although higher plasma VWF levels occur in blood group non-O vs O, there are no data to suggest that bleeding phenotype in those diagnosed with VWD differs by blood group, and blood group–specific reference ranges are not recommended.² 

Due to historical concerns of hormone-related fluctuations, the timing of sampling of plasma VWF levels was previously recorded in relation to menstruation. However, larger studies have demonstrated no difference in plasma VWF levels during the menstrual cycle, and timed testing is not recommended.^11,12  Of note, hormonal changes in pregnancy induce a 1.5-fold elevation in plasma VWF levels, hampering assessment for VWD.¹³  The role of exogenous estrogen is more complex. A literature review of ethinyl estradiol (EE) use (<50 µg/d) and plasma VWF levels identified 7 relevant studies.¹⁴  In 1 study, a mild reduction in plasma VWF:Ag levels was seen, but in all others, no significant changes were observed, as also seen in a more recent study.^14,15  Hormone replacement therapy may increase plasma VWF levels, and older data on high-dose EE indicated an elevation in plasma VWF levels with use.^16,17  Due to potential interference with plasma VWF levels, testing following high-dose EE therapy (>50 µg/d) should be interpreted with caution and plasma VWF levels rechecked once tapered.¹⁸  Since most modern COCPs include low-dose EE formulations, interruption is not necessary as it is unlikely to affect results and may worsen HMB and/or anemia.¹⁸ 

While plasma VWF:Ag or VWF:RCo >100 IU/dL has a negative predictive value >95% for VWD,¹⁹  some patients ultimately diagnosed with VWD may have initial test results within the normal range (>50 IU/dL).^19,20  In adolescents with acute HMB, if plasma FVIII:C was not elevated, only a minimal change in VWF:Ag and VWF:RCo levels was observed at retesting (2% and 1%, respectively), suggesting FVIII:C levels may help determine the need to retest.²⁰  In a small study of 22 iron-deficient women, retesting following iron supplementation identified type 1 VWD in 27% despite normal plasma VWF levels on initial testing.²¹  Due to the known fluctuations in plasma VWF, diagnosis in people with borderline plasma VWF levels is more challenging, often resulting in repeated plasma VWF testing.² 

This 14-year-old has a positive BS, borderline plasma VWF levels, and a negative family history. Once her anemia has corrected with alternate-day oral iron supplementation, we would repeat plasma VWF levels (including a VWF:GpIb assay) while continuing her COCP. Repeat testing identified low VWF levels (VWF:Ag 48 IU/dL, VWF:GpIbR 45 IU/dL, VWF:RCo 42 IU/dL) that were subsequently confirmed, and, as per recommendations for women and girls with bleeding disorders, she was linked into the combined hematology/gynecology clinic for ongoing care.²² 

A 19-year-old man with low VWF levels (VWF:Ag 46 IU/dL, VWF: RCo 44 IU/dL) reengaged with services for a wisdom tooth extraction. He was diagnosed at 6 years due to an aunt with type 1 VWD. He received prophylactic 1-deamino-8-D-arginine vasopressin (DDAVP) for tonsillectomy and had a mole removed with no treatment; no excess bleeding was observed with these procedures. On recent repeated testing, plasma VWF levels were >50 IU/dL (VWF:Ag 58 IU/dL and VWF:RCo 54 IU/dL). How will you manage this patient going forward?

When examining the impact of aging on plasma VWF levels, the focus is normally on the transition from middle to older age.^23-25  However, plasma VWF levels are dynamic throughout life, increasing from childhood to adulthood.^26,27  A recent study involving 1399 healthy children demonstrated mean plasma VWF:Ag of 82 IU/dL in 3- to 6-year-olds in comparison to 100 IU/dL in those aged 11 to 18 years.²⁷  Indeed, low VWF levels were seen in 5% of healthy 3- to 6-year-olds but did not correlate with postoperative bleeding.²⁷  An Italian longitudinal study of type 1 VWD found plasma VWF:RCo levels increased with age even in the youngest tercile (0.4-24.8 years; mean increase, 12 IU/dL) despite a relatively short duration of follow-up (mean, 3 years).²⁶  These age-related responses vary by subtype, with limited or no responses seen in older patients with type 2 and 3 VWD.²⁴ 

Plasma VWF levels 30 to 50 IU/dL, particularly as they approach 50 IU/dL, will overlap with the normal population, and diagnosis of VWD requires the presence of a bleeding phenotype.²  BATs should be used to aid phenotypic assessment, discriminating between abnormal and common bleeding symptoms.⁶  In our case, a diagnostic BAT was not performed, although medical notes document an uneventful delivery and childhood, no bleeding with vaccinations, and only short-lived epistaxis (<5 minutes), indicating a negative BS (ISTH-BAT 0); this was confirmed with the patient at review.

The inheritance patterns of low VWF (type 1 VWD with plasma VWF levels 30-50 IU/dL) are not clear, and pathogenic VWF mutations are infrequently identified. Tosetto et al²⁸  used a Bayesian approach to investigate the influence of AFM on the likelihood of diagnosis of type 1 VWD. A final odds ratio (OR) of ≥2.0 translated to a 66% positive predictive value of a diagnosis of VWD. Highest ORs were found in those with multiple affected first-degree family members. In contrast, for those who lacked AFM, an OR ≥2.0 often required plasma VWF:Ag levels <20 IU/dL. In those with plasma VWF:Ag levels in the 40- to 47-IU/dL range lacking AFM, a markedly elevated BS (condensed molecular and clinical markers for the diagnosis and management of type 1 [MCMDM-1] VWD >6; normal <4) was necessary to increase the final odds to ≥2.0.²⁸  As a result, the 2021 guidelines center the diagnosis of type 1 VWD with plasma VWF levels 30 to 50 IU/dL on bleeding phenotype rather than the presence of AFMs. In the case of our patient, only a second-degree relative was affected, and the patient's initial plasma VWF levels were mildly reduced. These features reduce the likelihood of VWD unless a significant bleeding phenotype was present, not suggested by the history.

The bleeding history since diagnosis may help guide future care, although only markedly elevated diagnostic BS (condensed MCMDM-1VWD scores >10) are predictive of future bleeding.²⁹  A structured bleeding history revealed no spontaneous or traumatic bleeding in childhood or with surgery. The procedure covered with DDAVP prophylaxis is not informative; however, a minor procedure (naevus removal) did not result in bleeding.

The cumulative data obtained from reassessment of personal bleeding phenotype, family history, and laboratory assays should be combined to help in the management of future care. In this instance, although diagnostic plasma VWF levels were <50 IU/dL, this was not coupled with either a personal or family history of bleeding. Since diagnosis, no bleeding events have occurred, and plasma VWF levels are now >50 IU/dL. If this individual were referred now, he would not meet criteria for VWD, but he did receive prophylaxis for some hemostatic challenges in the past (with no excess bleeding).

In this instance, we would suggest revisiting the diagnosis of type 1 VWD and discussing this with the patient. Key to this conversation is the patient's outlook on the diagnosis of VWD and impact on care. With some individuals, reconsidering the diagnosis will be welcomed as it will simplify future health care interactions. For others, it may precipitate an alteration in self-identity, and there may be considerable resistance to change. It is therefore essential to employ a shared decision-making approach.

For future procedures, we would advise expectant management with early intervention should bleeding occur. Inpatient procedures should ideally occur in a setting with hemostatic support rapidly available if necessary. If this patient completes multiple hemostatic challenges without bleeding, this would be reassuring that he lacks a bleeding tendency. Should excess bleeding occur, he should be reevaluated with consideration given to alternate diagnoses or bleeding disorder of unknown cause.

This case underscores the importance of careful consideration prior to assignation of a diagnosis of a bleeding disorder. For people with plasma VWF levels in the 30- to 50-IU/dL range, the cornerstone of diagnosis remains a positive bleeding phenotype. The implications of diagnosis should not be forgotten as they may be far reaching and not only medical in nature (eg, insurance weighting, familial counseling).

Would this approach differ if it was a 54-year-old man with type 1 VWD (baseline VWF:Ag 39 IU/dL, VWF:RCo 34 IU/dL) due to have an inguinal hernia repair? Recent levels are >50 IU/dL (VWF:Ag 58 IU/dL, VWF:RCo 54 IU/dL), but his diagnostic ISTH-BAT was 7 (epistaxis requiring cauterization, bruising, packing after dental extraction).

In adulthood, plasma VWF levels continue to rise, with a 1.5-fold increase seen between younger (<18 years) and older (>55 years) adults, with higher increments seen in the presence of comorbidities.^9,23,30-32  In those with type 1 VWD (<30 IU/dL), modest increments may be seen (median increase VWF:Ag 3 IU/dL/decade, VWF:Act 4 IU/dL/decade), but some studies have failed to show any improvement in levels.^24,26  More marked age-related increases are seen in those with low VWF. Plasma VWF:Ag and VWF:RCo increased by a mean of 1.9 IU/dL/y and 1.5 IU/dL/y, respectively, in adults followed with low VWF (n = 64), with plasma VWF levels ultimately >50 IU/dL in 29 of 64 patients.²⁵  Similar findings in type 1 VWD cohorts were reported by Rydz et al³³  (n = 31, VWF:Ag 30 IU/dL/decade, VWF:RCo 20 IU/dL/decade) and Abou-Ismail et al³⁴  (n = 126, VWF:Ag 2.4 IU/dL/y, VWF:RCo 1.4 IU/dL/y) (Table 1). As in the earlier case, this increase poses dilemmas to health care providers regarding management of future hemostatic challenges. It is also relevant to consider that techniques for the laboratory assessment of plasma VWF levels and their accuracy have improved over time. As before, we must approach this scenario by reviewing the original diagnosis, bleeding phenotype, and whether age-related increases in plasma VWF levels ameliorate phenotype in older adults.

Longitudinal patient cohort studies provide us with limited data focused on this issue. The WIN study compared 43 older (>65 years) and 344 younger patients with type 1 VWD (<30 IU/dL). Although plasma VWF:Ag levels increased in the older subgroup (median, 38 IU/dL [interquartile range, 24-53 IU/dL] vs 30 IU/dL [interquartile range, 19-46 IU/dL]), no difference in bleeding rate in the preceding 12 months was seen (30% vs 23%; P = .271). There was not, however, a specific focus on bleeding in those whose plasma VWF levels were >50 IU/dL. Examining perioperative outcomes from the LoVIC study (n = 60, 149 minor/dental and 11 major procedures), 6 experienced excess bleeding, all in minor/dental procedures in individuals with plasma VWF:Ag >50 IU/dL at the time of intervention.³⁵  A multicenter study following bleeding in patients with type 1 VWD found no significant difference between those with plasma VWF:RCo in the 30- to 50-IU/dL range and those who had an increase to >50 IU/dL.³  These data suggest that age-related increases in plasma VWF levels in adults may not translate into improvements in bleeding phenotype.

Circumstances surrounding diagnosis may differ between children and adults with VWD. Children are commonly diagnosed on the basis of family screening, but adults may not be diagnosed until later in life following repeated bleeding episodes and/or hemostatic challenges.^36,37  This provides greater insight into phenotype, the foundation of diagnosis. With improved plasma VWF levels, the individual's previous bleeding history should not be discounted as the association between the increased VWF levels and bleeding symptoms is not established.²  Our patient has a positive BS, and despite plasma VWF levels that have increased from his baseline into the normal range, the normal perioperative risk assessment procedures should be employed.³⁸  We would provide him with prophylaxis for the upcoming hernia repair with tranexamic acid and postoperative monitoring. Although a single dose of DDAVP preoperatively (if available) would boost preoperative plasma VWF levels, there is no evidence to guide whether DDAVP is still required in such patients preoperatively or whether tranexamic acid would suffice. If DDAVP is considered for use, especially in older patients, physicians must first assess cardiovascular risk and ensure no contraindications to use exist.

The diagnosis of VWD is unambiguous in those with markedly reduced plasma VWF levels (<10 IU/dL) or reduced levels and a strong bleeding phenotype, but many referrals involve people with mild bleeding symptoms and borderline plasma VWF levels. Differentiation between those with no propensity to bleed and low VWF remains clinically challenging. The presence of AFM increases the likelihood of a diagnosis of VWD, assisting evaluation, but for type 1 VWD with levels 30 to 50 IU/dL, the diagnosis hinges on the bleeding phenotype. It is vital, therefore, for physicians to understand the optimal uses and limitations of BATs. As people with type 1 VWD age, varying increases in plasma VWF levels will be seen, complicating diagnostic assessment and management. The current application of a single plasma VWF threshold throughout life fails to account for these age-related changes. To understand the biological differences resulting in these changes and the impact on bleeding phenotype, longitudinal studies involving both adults and children with levels >30 and <30 IU/dL will be required.

Robert F. Sidonio has served on advisory boards for Octapharma, Takeda, Guardian Therapeutics, Star Therepeutics, Genentech, Sanofi, Pfizer, Uniqure, Biomarin, Novo Nordisk, and Hema Biologics. He has also received investigator-initiated funding from Takeda, Octapharma, and Genentech.

Michelle Lavin has served on an advisory board for CSL Behring and as a consultant for Sobi, CSL Behring, and Band Therapeutics. She has also received speaker fees from Pfizer, with indirect funding for development of educational content from Takeda.

Robert F. Sidonio: nothing to disclose.

Michelle Lavin: nothing to disclose.