A diagnosis of von Willebrand disease despite normal test results for factor VIII and von Willebrand factor antigen and activity

A 32 year-old female with personal history of bleeding presents for clinical evaluation of a potential bleeding disorder. The issue of a bleeding diathesis was first raised four years ago when she developed extensive bruising/haematoma of her anterior thighs (25 × 10 cm), following a short period of rhythmic tapping in a dance class. The sole family history of bleeding then was her mother who also experienced “easy bruising”. This initial case presentation represents a familiar scenario for hematologists working with bleeding disorders. The patient presented with cutaneous bleeding following minor trauma, and a history of minor bleeding with an ISTH BAT (International Society on Thrombosis and Haemostasis Bleeding Assessment Tool) score of 6 (abnormal is ≥6 in adult females).1 Family history is similarly inconclusive, although a congenital bleeding disorder passed maternally cannot be excluded. At this stage, some basic laboratory testing is indicated. There are several plausible conditions to assess/exclude, although it is usual to start with simple tests, and initially evaluate for common bleeding disorders.2, 3 Table 1 identifies results from the initial round of tests. The patient presented with a mixture of normal and “abnormal” (flagged) results, although these did not provide satisfactory resolution to the plausible bleeding diathesis. The white cell count (WCC) was raised, comprising raised neutrophils and lymphocytes, but hemoglobin and platelet counts were normal, as were other parameters. As the leucocytosis resolved promptly, this was considered due to an intercurrent (upper respiratory tract) infection. The prothrombin time (PT) and activated partial thromboplastin time (APTT) were normal. Assessment of platelet function using the platelet function analyser (PFA-100) was normal, as was assessment by light transmission aggregometry (LTA) using several agonists (collagen, adenosine diphosphate [ADP], ristocetin, arachidonic acid, epinephrine). Testing by ristocetin included low dose testing (no response). Evidence of a biphasic aggregation was provided by both ADP and epinephrine. Assessment of von Willebrand factor antigen (VWF:Ag) and activity using several methods were all within respective reference intervals. Interestingly, factor VIII (FVIII) activity was raised. The initial test patterns observed appeared largely non-diagnostic. The VWF tests, all within normal reference ranges, seemed to discount von Willebrand disease (VWD), representing the most common inherited bleeding disorder, especially in females.4, 5 Platelet function tests6, 7 seemed to discount a platelet function defect. The normal PFA-100 further suggested exclusion of VWD, given its high negative prediction capacity (especially certain subtypes).7, 8 The raised FVIII was interesting, although this could (consistent with raised WCC) simply represent an acute phase elevation related to infection, although CRP (C-reactive protein) levels were noted to be normal. In any case, repeat laboratory evaluation is always warranted to confirm findings, due to both preanalytical issues,9 and intra-individual variation.10 This was arranged and repeat results are posted in Table 1. Repeat testing was similar to that of initial testing, although WCC count normalised. Notably, concurrent infection was not identified at second presentation. Also notable was platelet count (again normal), discounting thrombocytopenia, representing another common explanation for bleeding. The VWF tests were again within respective normal reference intervals, but FVIII remained high. PFA-100 testing was again normal. The repeated normal platelet count was reassuring, excluded thrombocytopenia and also potentially type 2B VWD or platelet type (PT-)VWD, both generally associated with (mild) thrombocytopenia.11, 12 The 2B/PT-VWD exclusion was also consistent with normal aggregation to ristocetin (ie, no low dose response identified by ristocetin induced platelet agglutination [RIPA] assay).13 The VWF test results (all analytes within normal range) continued to suggest an absence of VWD. Nevertheless, the patient was identified to be on an oral contraceptive pill (OCP) during prior testing, and since hormonal treatment may affect hemostasis tests,14 a third round of testing (3rd presentation) was arranged (off OCP), with results also identified in Table 1. These were essentially consistent with previous testing, although FVIII was now normal. Thus, although OCP treatment was not thought to have extensively affected previous results, an effect on FVIII could not be discounted. But, as CRP was normal, the elevated FVIII might not reflect an acute phase response. It was decided to manage the patient conservatively, and await further developments. As part of a patient follow up, testing was repeated some 3.5 years after the initial investigation. This was in part due to recommencement of the OCP because of continued menorrhagia (results detailed in Tables 2 and 3 ; 4th presentation). A marginally raised WCC count, including raised neutrophil count, was again observed. Again, FVIII was raised (Table 2 ). The VWF tests were again within their normal reference intervals, except for collagen binding (VWF:CB), where a very low value of 10 U/dL was obtained (Table 3 ). This greatly surprised the testing laboratory, and accordingly was repeated on the same sample using the same methodology, and subsequently confirmed (11 U/dL; Table 3 ). Given the unusual finding, this being inconsistent with previous VWF:CB results, the laboratory suspected a methodological issue. It was arranged for an additional sample to be collected (5th sample) and tested at two other sites, with results also identified in Table 3. The VWF:CB was similarly low when assessed using the same methodology, but within the normal reference range using a different methodology (similar to the original methodology). The slightly raised WCC was again likely explained by another recent infection, and the raised FVIII potentially influenced by recommencement of the OCP. The most striking result, however, was the variable VWF:CB values observed on different occasions and in different laboratories, with this related to different methodologies. Further details around the variable VWF:CB data and other VWF test methodologies are summarised in Table 3. This table has now also standardised the way that ratios of VWF Activity/Ag are expressed, and also applied the internationally recommended cut-off of 0.6 to these ratios.15 Essentially, the classically applied VWF:CB methodology was a commercial enzyme linked immunosorbent assay (ELISA), with the collagen source being not defined by the manufacturer (Life Therapeutics). This methodology invariably gave results within the normal reference interval for that assay, both as historically tested (Table 1) or as tested during subsequent investigation (Table 3; Lab B). Instead, the alternate VWF:CB methodology more recently applied, representing a newer chemiluminescence-based methodology (ACL AcuStar; using a collagen type III peptide), invariably gave very low VWF:CB values, irrespective of where the test was performed (Tables 3; Labs A, C). The chemiluminescence-based VWF:CB method was also associated with corresponding very low VWF:CB/Ag ratios, irrespective of VWF:Ag assay used. The provisional diagnosis now became type 2M VWD, of a “specific” or “unusual” collagen binding type. In order to help confirm this diagnosis, additional testing was arranged, specifically VWF multimer analysis and genetic testing. Family studies were also planned. VWF:multimer analysis did not identify any altered multimer structure (as would be expected for 2M VWD16; Figure 1). Molecular analysis by next generation gene sequencing analysis of VWF, F8, GPIBA and GPIBB detected a variant in VWF at exon 31 (c.5347 T > G) that would be classified according to the American College of Medical Genetics (ACMG) guidelines17 as a “variant of uncertain clinical significance”. This gene variation leads to a VWF protein change of p.Ser1783Ala, and is expected to affect collagen binding according to previous literature.18, 19 Unfortunately, although family studies were planned, it was later discovered that the patient was adopted. Although the biological mother was identified, she refused testing; the biological father was not available for testing. Two step-sisters on the maternal side were also identified, although neither have an identifiable history of significant bleeding, and neither is yet willing to be tested. Thus, family studies are not yet possible. VWD is well-identified as the most common congenital bleeding disorder, and arises from deficiency and/or defect in plasma VWF. Three main VWD types are recognised (1,2,3), with a total of six types (1, 2A, 2B, 2M, 2N, 3) identified according to current ISTH classification.20 Types 1 and 3 identify quantitative deficiency of VWF, respectively partial or (virtually) total; importantly, any VWF present is functionally normal, albeit deficient in quantity. Type 2 VWD defines qualitative defects of VWF, although VWF levels are also quantitatively low in many patients. Type 2A VWD identifies patients with a specific loss of high molecular weight (HMW) VWF, otherwise representing VWF forms that have greatest specific adhesive activity. Type 2A VWD patients suffer from loss of VWF activity binding to both platelets (primarily via glycoprotein Ib; GPIb) and subendothelial matrix components (predominantly collagen). Type 2B VWD identifies patients with hyper-functional VWF, which spontaneously binds platelets, thereby causing clearance of (predominantly HMW) VWF and platelets from circulation, and often a (mild) thrombocytopenia. Type 2N VWD reflects a loss of VWF - FVIII binding; thus, increasing FVIII susceptibility to degradation, and reducing plasma FVIII levels. Type 2M VWD reflects a heterogeneous group with defective VWF not associated with loss of HMW VWF. In addition to clinical/physical assessment of patients, VWD diagnosis is aided (or alternatively discounted) by laboratory testing.3 A standard test panel recommended by most guidelines15, 20, 21 comprises at least FVIII, VWF:Ag and VWF:RCo. FVIII testing is important since one major function of VWF is to bind and protect FVIII from degradation. Thus, FVIII is often low in VWD, and may also contribute to any bleeding diathesis (especially types 3, 2N, and severe type 1 VWD). VWF:Ag testing is important due to complexity and heterogeneity of VWD and permits quantifying absolute levels of (total) VWF protein. Note, VWF:RCo represents the historical VWF “activity” assay, reflecting VWF - GPIb binding. Additional assays may be used in some countries or laboratories. VWF:CB is a functional assay identifying binding of VWF to subendothelial matrix collagen. VWF:CB performance is recommended by the UK Haemophilia Doctors guideline,15 and also mentioned in others.20, 21 In vivo, collagen binding permits anchorage of platelets to damaged endothelium, enabling VWF to act as an adhesive bridge to capture platelets. Assessment of VWF multimers is also useful, particularly to help identify loss of HMW VWF (ie, types 2A or 2B VWD), and potentially discriminate these from types without such loss (eg, type 1 or 2M VWD).15, 20, 21 However, laboratories with comprehensive test panels that combine GPIb binding22, 23 with VWF:CB24, 25 may be able to predict VWF multimer patterns based on these results.16, 26 Thus, low activity values for both test classes, as well as low activity/Ag ratios for both classes, likely represents loss of HMW VWF. In contrast, if only one class of activity test is low (ie, GPIb binding or VWF:CB), as associated to only one low activity/Ag ratio (but not both), then this pattern is “incompatible” with loss of HMW VWF and therefore likely represents type 2M VWD.3, 27 Genetic testing is a final stage in diagnosis and classification, although emergence of next generation sequencing may alter its future positioning.28 Nevertheless, genetic testing is not mandated, nor required for a diagnosis of VWD, which currently relies on evidence of a bleeding history and results of phenotypic testing (this is primarily due to cost and complexity of genetic testing, and because many VWD cases are never identified to have a VWF “mutation”).15, 20, 21 Thus, for current VWD diagnostics, genetic testing is generally applied to a minority of cases (primarily types 2 and 3 VWD).29, 30 The ability of laboratories and clinicians to identify/exclude VWD is dependent not only on the number of tests employed in the VWD testing panel, but also type and combination of tests, and recognition of test patterns.31 For example, VWF:CB is performed by most VWF test laboratories in Australia,32 a high proportion in Europe, but few laboratories in the USA.33 This in part relates to greater availability of regulatory approved/cleared VWF:CB assays in Australia and Europe, vs absence in the USA.34 In contrast, VWF multimer testing is performed by more laboratories in the USA than in Australia.32, 33 In part, this is since laboratories have greater need for multimer testing in the absence of a VWF:CB, since (as explained above) the combination of VWF:CB and GPIb binding assay testing can help predict the multimer pattern. On the other hand, multimer assays are often performed poorly or misinterpreted, and can themselves lead to high levels of VWD misdiagnoses.33 Our patient was unusual, and her initial diagnosis as 2M VWD missed, for several reasons. First and foremost, initial testing for all VWF analytes identified values within their respective normal reference intervals, during both initial and repeat testing (Table 1). This testing included FVIII, VWF:Ag (by Stago LiaTest), VWF:RCo (Siemens method initially), VWF:CB (Life Therapeutics) and also another VWF “activity” assay - this being the Siemens Innovance VWF Ac assay23, 35 (designated VWF:GPIbM by the ISTH).3, 36 These tests were performed three times, both on and off OCP, with reasonable internal consistency. Although FVIII was higher when the patient was on OCP than off OCP, VWF test results seemed to discount VWD. Performance of multimers at that time would not have changed this outcome or alerted to a case of VWD, since there was (as expected in 2M VWD, and also on the basis of normal VWF:RCo and VWF:GPIbM and VWF GPIb-binding/Ag ratios) no loss of HMW VWF, when eventually performed (Figure 1). However, perhaps under-recognised during the initial investigation was the potentially raised VWF:Ag/CB ratio obtained on all test occasions (Table 1; since the laboratory never defined a standard cut-off value for “abnormality” of this ratio). Indeed, it was not until VWF:CB testing was redone using a newer and different chemiluminescence methodology, that VWD was finally identified in this patient (as type 2M). This form of VWD is both under-recognised and misidentified as either type 2A or type 1 VWD, depending on the case and tests employed.27, 37 Thus, if the 2M VWD defect affects only GPIb binding, and thus only VWF:RCo (or alternate GPIb-binding tests), but not collagen binding (thereby yielding normal VWF:CB), then most laboratories will assume loss of HMW VWF and type 2A VWD, unless clarification is actually obtained by multimer analysis (no HMW loss). Alternatively, or additionally, VWF:CB testing would assist in this diagnosis.27, 37, 38 This was well evidenced recently, where two GPIb binding defect type 2M VWD cases were tested by many laboratories via external quality assessment.27 One sample (median 49 U/dL VWF:Ag), was more often misidentified as type 2A/2B (46.7%) than 2M (34.8%) VWD. Another 2M sample (median 189 U/dL VWF:Ag), was instead often misidentified as being normal (non-VWD) (36.4%), with identifications of type 2A/2B VWD (13.6%) also represented. Laboratories that included VWF:CB testing were much less likely to report errors in diagnosis. However, our current case is not a type 2M GPIb binding defect (2MGPIb), but rather type 2M VWD of collagen binding type (2MCB), which in total represent fewer identifications of 2M than those of 2MGPIb. Indeed, there have been only 11 type 2MCB cases identified in the literature.38-40 Although it is likely that there are numerically fewer cases of type 2MCB than 2MGPIb, it should also be acknowledged that under-recognition is also occurring, as cases will not be identified unless VWF:CB is performed. Indeed, our case is even more unique, since VWF:CB testing was initially performed, but the original methodology did not well identify the collagen binding defect, yielding “normal” values for VWF:CB. Albeit in hindsight concurrently associated with increased VWF:Ag/CB (or reduced VWF:CB/Ag) ratios, with this pattern not clearly indicative of type 2MCB at that time. Indeed, this collagen binding defect only became clearly identified when a new VWF:CB methodology, based on chemiluminescence, was applied. This differs from classical VWF:CB testing, which is mostly performed by ELISA using type I, type III or type I/III collagen mixtures.37, 41 The chemiluminescence VWF:CB procedure instead uses a type III collagen peptide to measure collagen binding.42 This methodology appears more sensitive to type 2A and 2M (especially 2MGPIb) VWD than ELISA assays,42 which again reflects on the heterogeneity of VWD, including type 2M VWD, counting 2MCB. The chemiluminescence VWF:CB procedure is not yet available in the USA, but is widely available in Australia and in Europe. It is nonetheless also assuring, however, that US workers are active in the arena of 2M VWD, as VWF:CB, in and/or Our case has a VWF genetic variation (c.5347 T > to a change in VWF that has only been in a The female was following gave a history of and her bleeding score was the of all cases in that Our patient also has significant but has not as yet experienced or The affected and and a bleeding score of The VWF:Ag U/dL; VWF:RCo and multimer analysis were all normal, as were using the The normal for these cases our patient may also reflect on the collagen type being type according to the manufacturer VWF:CB was reduced in prior cases with both collagen type 27 and III 19 in VWF:CB/Ag ratios of to Our case plasma levels of VWF for all VWF around those of the and accordingly all VWF test results were including VWF:CB as initially performed. The for the may to as the cases were group our case is group A, and many studies have identified that group have on less VWF than This also more patients with group being identified with VWD to other simply because there is a higher for to VWF levels the normal reference some group may have a diagnosis of VWD because VWF levels are elevated due to and a potential genetic The genetic change identified in our case and has studies to further identify the specific of collagen The protein structure and of VWF with collagen has also been and the identified as being to the binding studies the in the where this genetic change have to be that otherwise leads to of collagen Although the is expected to with is its effect on collagen binding may via of the as a As and only in of a group by in the side the original this even minor in of the cannot be In addition to being also identified on the VWF gene this genetic change was further by studies using of and the of the T > change in both collagen binding and also Nevertheless, were by that results not reflect the disease in In report a case of VWD, with a specific defect of collagen binding, that the of type 2MCB. The case presented with a BAT score of on the low of but with significant bleeding history including The case was eventually as in VWF:CB/Ag ratio using all but VWF:CB only the reference interval when performed by a new chemiluminescence based Nevertheless, it is that other cases of 2M VWD may be detected by other VWF:CB Thus, one method another based on this case may be should be that it is for different methods to test negative (or potentially even with certain patients. For example, this can with some (eg, agglutination yielding normal or elevated values of VWF due to such as and thus potentially negative for VWD. Our case also presented several including elevated WCC most likely due to intercurrent and elevated FVIII associated with OCP acute and/or reflecting group A related The case has a genetic variation that has only been on one other and thus likely reflects a form of VWD. However, the ability of laboratories to identify such cases of sensitive VWF:CB and recognition of low VWF:CB/Ag Additional cases may be but may due to performance of VWF:CB or of certain VWF:CB or under-recognition of VWF:CB/Ag ratios should VWF:CB and VWF:Ag levels be normal Our patient will to be family studies will be and additional to of VWF during and of will be A of may be indicated.