A, Transesophageal echocardiography revealed a torn tissue prosthetic cusp and severe mitral regurgitation. B, von Willebrand factor multimers revealed loss of highest-molecular-weight multimers (top gel, with left arrow at band 15, middle arrow at band 10, and right arrow at band 2). C, The patient underwent transfemoral valve-in-valve replacement with a 29-mm valve (SAPIEN XT; Edwards Lifesciences Corporation). High-molecular-weight multimer loss resolved by both dichotomous visual inspection and quantitative multimer ratio of bands greater than 15 divided by bands 2 through 15 (from 0.58 to 0.85). There was also improvement in Platelet Function Analyzer 100 (Siemens USA) closure time (from 143 to 77 seconds) (normal is <121 seconds) and in von Willebrand factor activity to antigen ratio (from 0.67 to 0.77).
Native aortic regurgitant volume correlation with Platelet Function Analyzer 100 (Siemens USA)–collagen adenosine diphosphate (PFA-CADP) (r = 0.74; P < .001).
For purposes of showing the relative severity of von Willebrand factor multimer disruption, data on healthy individuals, patients with LVADs, and patients with AS are included. Some of the data on healthy individuals and those with AVR and MR have been previously published by Blackshear et al.4,11,15 AR indicates aortic regurgitation; AS, aortic stenosis; AVR, aortic valve replacement; LVAD, left ventricular assist device; MR, mitral regurgitation; MV, mitral valve; MVR, mitral valve replacement; SAVR, surgical aortic valve replacement; and TAVR, transcatheter aortic valve replacement.
eFigure. Function of AdamTS13
eTable. Type of Valve Lesion
Customize your JAMA Network experience by selecting one or more topics from the list below.
Blackshear JL, McRee CW, Safford RE, et al. von Willebrand Factor Abnormalities and Heyde Syndrome in Dysfunctional Heart Valve Prostheses. JAMA Cardiol. 2016;1(2):198–204. doi:10.1001/jamacardio.2016.0075
Limited data suggest that von Willebrand factor (VWF) abnormalities may accompany the high-shear state associated with prosthetic valve dysfunction. If true, laboratory testing could add value in quantifying prosthesis dysfunction and could suggest a pathophysiological explanation for acquired bleeding in some patients.
To determine whether dysfunctional valve prostheses are associated with VWF abnormalities compared with normally functioning valve prostheses, to identify the severity of the VWF abnormality relative to other conditions, and to describe associated bleeding and the occurrence of gastrointestinal angiodysplasia.
Design, Setting, and Participants
Cohort study in a multispecialty practice setting from August 2010 through November 2015. To assess the severity of VWF dysfunction, data were compared with those from previously reported healthy controls and patients with aortic stenosis, mitral regurgitation, and left ventricular assist devices. Patients underwent assessment of multiple VWF laboratory tests and echocardiography.
Main Outcomes and Measures
Loss of high-molecular-weight multimers of VWF.
A total of 136 patients were included in this study. During the study period, we assessed 26 patients with normally functioning surgical or transcatheter aortic valve replacement, 24 patients with dysfunctional aortic valve replacement, 36 patients with normally functioning mitral valve replacement or repair, 19 patients with dysfunctional mitral valve replacement or repair, and 31 patients with native aortic regurgitation without coexisting aortic stenosis. von Willebrand factor multimers were abnormal in 1 of 26 normal aortic valve replacements and in 2 of 36 normal mitral valve replacements or repairs but were abnormal in 20 of 24 dysfunctional aortic valve replacements and in 14 of 19 dysfunctional mitral valve replacements or repairs (P < .001 for both). Normal aortic valve replacement also had a higher VWF activity to antigen ratio, mean (range) 0.94 (0.84-0.99) compared to dysfunctional aortic valve replacement, 0.78 (0.73-0.87), P < .001, as did normal mitral valve replacement or repair, 0.90 (0.86-0.93) compared to dysfunctional mitral valve replacement or repair, 0.78 (0.70-0.90), P = .005. Platelet function analyzer closure times were lower with normal aortic valve replacement, mean (range) 92 (82-112) seconds compared to dysfunctional aortic valve replacement, 139 (122-177) seconds, P < .001, and also in normally functioning mitral valve replacement or repair, 85 (74-96) seconds compared to dysfunctional mitral valve replacement or repair, 143 (128-192) seconds, P < .001. Gastrointestinal bleeding was noted in 6 of 24 patients with aortic prosthesis dysfunction and in 5 of 19 patients with mitral prosthesis/repair dysfunction and was associated with a lower normalized VWF multimer ratio than in patients without bleeding. Gastrointestinal angiodysplasia was noted in 5 of 6 bleeding patients with dysfunctional aortic prostheses and in 3 of 5 bleeding patients with dysfunctional mitral prostheses/repair.
Conclusions and Relevance
Acquired abnormalities of VWF multimers are associated with aortic and mitral prosthesis dysfunction, with occasional gastrointestinal bleeding and gastrointestinal angiodysplasia. Quantitative VWF tests may provide adjunctive value in the difficult assessment of prosthetic valve dysfunction.
Heyde syndrome, the association of severe aortic stenosis (AS) and gastrointestinal bleeding due to angiodysplasia, has been linked to intravascular shear–induced loss of the highest-molecular-weight multimers of von Willebrand factor (VWF) (eFigure in the Supplement),1 but a similar constellation of findings has been seen in aortic regurgitation (AR) and in mitral insufficiency.2-4 Persistence of or recurrence of VWF abnormalities after valve replacement or repair may signal patient-prosthesis mismatch.3 Limited additional data suggest that VWF abnormalities may also accompany the high-shear state associated with prosthetic valve dysfunction, including 5 cases of mitral paraprosthetic leak reported by Pérez-Rodríguez et al,5 2 cases of transcatheter aortic valve replacement (TAVR) moderate to severe paraprosthetic leak reported by Spangenberg et al,6 and 4 cases of post-TAVR moderate to severe paraprosthetic leak reported by Van Belle et al.7 Findings from these studies, most notably the study by Van Belle et al,7 suggest a wider usefulness of VWF activity tests as adjuncts to imaging for early and late prosthesis assessment in both surgical and transcatheter valves. Their study found that patients with severe AS undergoing TAVR had normalization of VWF multimers 180 minutes after the procedure, while no recovery of VWF multimers was seen after balloon aortic valvotomy, in which peak aortic velocity declined only from a mean of 4.5 to 3.9 m/s.7 Finally, a new concern is prosthetic valve dysfunction related to reduced bioprosthetic leaflet motion, which is associated with subclinical thrombosis.8
Persistently abnormal VWF function in dysfunctional surgical or transcatheter prostheses could be associated with gastrointestinal bleeding over time. Indeed, Généreux and colleagues9 described major late bleeding complications in 5.9% of patients after TAVR. Gastrointestinal bleeding was the most frequent type of bleeding, and moderate to severe paravalvular leak was an independent predictor of bleeding. Herein, we report prospectively studied patients with late dysfunctional mitral and aortic prostheses and TAVR compared with important reference groups and include data on VWF testing, bleeding, and angiodysplasia.
Question Can von Willebrand factor (VWF) functional laboratory tests distinguish normal cardiac valve prosthesis function from prosthesis dysfunction?
Findings In a cohort study of 31 patients with mild, moderate, or severe native AR, VWF multimers were abnormal in 37.5% (3 of 8) of mild to moderate and 76.9% (10 of 13) of moderate to severe AR compared with 0% (0 of 10) of mild AR—a significant difference. Among 62 patients with normal cardiac prostheses, VWF multimers were abnormal in 4.8% (3 of 62) compared with 79.1% (34 of 43) of patients with varying types of at least moderate prosthesis dysfunction—also a significant difference—and acquired bleeding was present in 9.5% (2 of 21) of patients with at least moderate native AR and 25.6% (11 of 43) of patients with at least moderate prosthesis dysfunction.
Meaning Native AR and left-sided cardiac prosthetic valve dysfunction may be associated with acquired von Willebrand syndrome.
This cohort study in a multispecialty practice setting was conducted from August 2010 through November 2015. Beginning in 2010, patients with AS or normal surgical aortic valve replacement (SAVR) or mitral valve replacement were recruited to provide research blood samples. In 2012, the protocol was modified to allow testing of the hypotheses that AR and prosthesis dysfunction are associated with VWF abnormalities. Groups of patients were recruited with AR, native mitral regurgitation (MR) before and after surgery, dysfunctional aortic or mitral valve replacement or repair (MVR/rep), and TAVR, as well as control patients and patients with left ventricular assist devices (LVADs). The protocol allowed inclusion of data from patients who had valve disorders and had VWF laboratory testing performed for the clinical indication of acquired bleeding. The protocol was approved by the Mayo Foundation Institutional Review Board, and written informed consent was obtained. As a referral center for advanced endoscopy, including a high volume of double-balloon enteroscopy, referral bias was expected to result in a higher bleeding prevalence than would be anticipated in general clinical practice. Bleeding histories and echocardiographic data were recorded. This study was registered at ClinicalTrials.gov as NCT01334801.
Among patients with native AR, aortic valve anatomy was assessed, and patients with more than mild associated AS (peak aortic velocity ≥3 m/s) were excluded. Aortic regurgitant volume (RV) was calculated by the Doppler volumetric method, and AR severity was defined using RV criteria, according to guidelines,10 as mild (RV <30 mL), moderate (RV 30-45 mL), or moderate to severe (RV >45 mL). Assessments of AS and MR severity have been previously described.4,11 A dysfunctional prosthesis was defined as at least moderate regurgitation or stenosis by the interpreting clinical echocardiographer or cardiologist (J.L.B. and R.E.S.).
von Willebrand factor antigen, VWF activity, and qualitative loss of high-molecular-weight multimers were recorded as previously described.11 Qualitative loss vs nonloss of VWF high-molecular-weight multimers was the primary study end point in testing the study hypotheses, and it was determined by visual inspection of the electrophoretic gel compared with pooled normal plasma by physicians in the Mayo Clinic Special Coagulation Laboratory (D.C. and other nonauthors). For secondary analysis and comparison of types of valvular abnormalities, assessment of quantitative loss of VWF ratios of multimers exceeding band 15 divided by bands 2 through 15 was performed using patients’ citrated frozen plasma samples as previously described.11 This technique is similar to the quantitative method described by Spangenberg and colleagues,6 although our method eliminates band 1 from analysis because it is known to have the highest coefficient of variation of all the multimer bands12 and is contaminated with proteins other than VWF. To alleviate the significant gel-gel variation, we normalized the multimer ratio of patient plasma to the same gel control sample as previously done by others,2,7 which results in healthy individuals clustering around unity and with progressively more severe multimer loss reflected by a lower value (see the case example in Figure 1). Platelet Function Analyzer 100 (Siemens USA) adenosine diphosphate cartridge closure time (hereafter PFA-100) was assessed using citrated whole-blood samples. Besides its ability to detect platelet disorders, PFA-100 is a sensitive screening tool for congenital and acquired von Willebrand syndromes.13,14 The test measures the time required for fresh whole blood to clog an orifice on a cartridge coated with platelet activators, collagen, and adenosine diphosphate. Prolonged PFA-100 closure time is a sensitive indictor for acquired von Willebrand syndrome from cardiac disorders.3,7,11 PFA-100 adenosine plus collagen is not sensitive to aspirin or vitamin K antagonists, which are the most common anticoagulants used in our study population. However, it may be affected by hemolysis, severe anemia, and thrombocytopenia. The normal reference range for closure time in our laboratory is 57 to 121 seconds.
For this analysis, we included updated cohorts of previously reported patients with normally functioning SAVR11 or MVR/rep (see Table 1 and Table 2 footnotes).4 All patients with native AR and prosthesis dysfunction have not previously been reported. Because paravalvular regurgitation is the most prevalent type of prosthesis dysfunction and because the effect of native AR on VWF tests had only received limited prior study,2 we first sought to test whether native AR was associated with abnormalities of VWF function. We compared RV in patients with native AR with the primary outcome variable (abnormal VWF multimers) with RV in patients with normal multimers using a Wilcoxon rank sum test. Next, we compared native AR RV with quantitative measures of VWF function, activity to antigen ratio PFA-100, VWF multimer ratio, and normalized VWF multimer ratio using Spearman rank correlation (ρ). The frequency of abnormal VWF multimers in dysfunctional SAVR or TAVR was compared with normally functioning SAVR or TAVR using the Fisher exact test. Next, we made similar comparisons of the above-noted quantitative measures of VWF in dysfunctional SAVR or TAVR vs normally functioning SAVR or TAVR using the Wilcoxon rank sum test.
The frequency of abnormal VWF multimers in dysfunctional MVR/rep was compared with normally functioning MVR/rep using the Fisher exact test. We made similar comparisons of quantitative measures of VWF in dysfunctional MVR/rep vs normally functioning MVR/rep using the Wilcoxon rank sum test. Finally, to illustrate the relative severity of disruption of VWF multimers, we created graphs of the mean (SD) normalized multimer ratio for native AR, normally functioning and dysfunctional SAVR or TAVR, and normally functioning and dysfunctional MVR/rep plotted alongside previously reported4,11,15 cohorts of healthy individuals, patients with LVADs, and patients with native AS and native MR.
A total of 136 patients were included in this study. In Tables 1 and 2, basic demographic data and data regarding VWF function and brain natriuretic peptide are reported for patients with native AR and normal and dysfunctional prostheses, and the valve type and type of prosthetic dysfunction are listed in the eTable in the Supplement. Patients with native AR did not manifest AS, with a mean (SD) peak velocity of 1.9 (0.5) m/s. In the patients with dysfunctional prostheses, anemia (hemoglobin level <10 g/dL) was present in 21 of 42, and definite hemolysis (lactate dehydrogenase level >460 U/L plus anemia) was present in 5 of 36 in whom lactate dehydrogenase was measured (to convert hemoglobin level to grams per liter, multiply by 10.0; lactate dehydrogenase level to microkatals per liter, multiply by 0.0167). As shown in Table 1, patients with mild native AR had no loss of high-molecular-weight VWF multimers, but abnormal VWF multimers were present in 76.9% (10 of 13) of patients with moderate to severe AR. Among patients with AR, RV was higher in those with abnormal VWF multimers compared with those with normal multimers (P < .001 by the Wilcoxon rank sum test). PFA-100 had the strongest correlation with RV (ρ = 0.74, P < .001) (Figure 2), followed by normalized multimer ratio (ρ = −0.50, P = .004). There was no evidence of an association of VWF activity to antigen ratio with RV (ρ = −0.29, P = .11).
As summarized in Tables 1 and 2, abnormal VWF multimers were far more common in patients with dysfunctional aortic prostheses compared with those with normal prostheses, which was reflected by quantitative normalized multimer analysis, PFA-100, and activity to antigen ratio (P < .001 for all). Similar findings were observed for dysfunctional vs normal mitral prostheses (P < .001 for all). Brain natriuretic peptide was significantly higher in patients with dysfunctional aortic prostheses compared with patients with normal prostheses, and nonstatistically significant similar trends were seen for mitral valves.
Gastrointestinal bleeding was noted in 2 of 21 patients with at least moderate AR, in 6 of 24 patients with dysfunctional SAVR or TAVR, and in 5 of 19 patients with dysfunctional MVR/rep, as well as 1 patient each with excessive procedural bleeding and subdural hematoma with minimal trauma. Patients with dysfunctional prostheses and bleeding had a lower normalized multimer ratio vs nonbleeding patients with dysfunctional prostheses (mean [SD], 0.59 [0.13] vs 0.72 [0.18]; P < .03). Among the 6 patients with aortic valve prosthetic dysfunction and gastrointestinal bleeding, angiodysplasia was documented in 5 of them, while 1 had a single point of bleeding in the jejunum. One patient each with mechanical thrombosis and late bioprosthetic severe regurgitation underwent a second surgery, with recovery from bleeding. The other 4 patients had severe aortic paraprosthetic regurgitation, of whom 2 have died and 2 underwent percutaneous plugging with remission from bleeding. Among the 5 patients with bleeding associated with dysfunctional MVR/rep, 3 had documented angiodysplasia and 2 did not undergo endoscopy. One patient with prosthesis-associated left ventricular outflow obstruction was successfully managed by replacement of a digoxin regimen with a β-blocker. A patient with mitral mechanical paraprosthetic leak was treated surgically and recovered. Two patients underwent percutaneous plugging, 1 with remission of bleeding and 1 with continuation of bleeding. One elderly patient with severe MR after mitral valve repair has been treated with endoscopic photocoagulation and iron supplementation.
In Figure 3A, we show the mean (SD) normalized VWF multimer ratios in patients with AR, AS, normal prostheses, and dysfunctional prostheses, as well as the reference groups of healthy individuals and patients with LVADs. In a similar fashion in Figure 3B, we show the mean (SD) quantitative normalized VWF multimer ratios in 53 patients with varying degrees of native MR and in patients with normally functioning MVR/rep, dysfunctional MVR/rep, and LVADs.
Except for 9 patients with surgically severe AR reported by Weinstein and colleagues,2 the association of native AR with VWF abnormalities has not been previously studied. Our data suggest that, in native AR (as is the case in mild, moderate, and severe AS or MR), abnormal results of 3 independent tests of VWF function worsen with increasing lesion severity, but the degree of impairment is less than is seen in AS or MR, possibly due to the lower driving pressure in diastole vs systole. In patients with prosthetic valve dysfunction, most of whom had paravalvular or late valvular insufficiency, we also found prevalent abnormalities of VWF function and the occurrence of gastrointestinal bleeding and angiodysplasia, as has been found in other high-shear states associated with acquired von Willebrand syndrome.16 Normalized VWF multimer ratios revealed mild to moderate abnormalities associated with normal SAVR or TAVR, as well as more severe abnormalities of dysfunctional prostheses, with some values overlapping those of severe AS, MR, and LVAD. These findings have 2 potential clinical implications.
First, 1 or more tests of VWF activity may be viewed as biomarkers of AR severity (or “biological sensors of blood flow,” as stated by Van Belle and colleagues7) and of prosthesis function, as was recently suggested by Spangenberg and colleagues,6 who (along with other authors) confirm the prevalent abnormalities of VWF activity in AS and correction with aortic valve replacement. Among post-TAVR patients, both of these studies6,7 suggested a low prevalence of VWF abnormalities after TAVR, while the study by Vincentelli and colleagues3 among post-SAVR patients found a late prevalence of patient-prosthesis mismatch of 26%, as well as abnormal values of VWF multimers in 74% and of PFA-100 measurements in 66%. Prior work suggesting that the hemodynamics of TAVR are superior to those of SAVR17 is concordant with these observations that VWF function is less frequently abnormal in post-TAVR patients vs post-SAVR patients. The data by Van Belle and colleagues7 correlating the severity of post-TAVR or balloon aortic valvotomy gradient to VWF multimer normalized ratio (r = −0.68) confirm prior data suggesting a biomarker effect in AS.3,11 At present, guideline-sanctioned biomarker assessments for prosthetic valves include serial hemolysis-related tests by European authors.18 We can foresee that serial assessments of normalized multimer ratio and other VWF tests may also be clinically useful in all types of left-sided tissue valves to characterize the severity of valvular or paravalvular leak or to detect subclinical bioprosthetic thrombosis or late failure due either to stenosis or tear with regurgitation.
A second potential clinical implication is the finding that patients with prosthesis dysfunction may have acquired severe bleeding, as was seen in 25.6% (11 of 43) of our patients with prosthesis dysfunction and in 9.5% (2 of 21) of our patients with moderate to severe native AR, which is a certain overestimate of true prevalence due to referral bias. Severe gastrointestinal bleeding is considered a major complication of congenital or acquired von Willebrand syndrome16 and, when defined by clinical and laboratory criteria, could constitute an indication for intervention. Laboratory confirmation of high-molecular-weight multimer loss in patients with severe bleeding and AS, hypertrophic cardiomyopathy, or MR provides impetus toward corrective surgery, which is expected to be curative of bleeding.4,19-22 A high percentage of our patients with prosthesis dysfunction were receiving anticoagulant therapy, and, in practice, a diagnosis of acquired von Willebrand syndrome might not be investigated in such patients when bleeding could be attributed to anticoagulant therapy. However, the presence of gastrointestinal angiodysplasia in patients with prosthesis dysfunction and abnormal VWF multimers points to acquired von Willebrand syndrome as the etiology of bleeding.16,23 Additional data are needed to determine whether percutaneous approaches, such as valve-in-valve replacement for late prosthesis failure (Figure 1), plugging to correct paravalvular leak, and percutaneous clipping of severe organic MR, correct loss of high-molecular-weight multimers and elicit durable remission from gastrointestinal bleeding.
Our study has several limitations. First, in the native AR population, only 4 patients had truly severe (>60 mL/beat stroke volume) AR. Second, the mechanisms of prosthetic valvular dysfunction are heterogeneous, and, although prosthetic valvular and paravalvular insufficiencies are the main lesions reported, the behavior of VWF multimers may differ by mechanism, and we were underpowered to assess this difference should one exist. Third, the number of patients with severe bleeding who underwent intervention is insufficient to provide clear guidance on when intervention is clearly indicated, and future investigations incorporating clinical data, echocardiography, VWF laboratory assessments, and endoscopic findings are needed.
These results provide support for the concept that AR and prosthetic valve dysfunction are associated with impairment of VWF function as well as that bleeding in the setting of dysfunctional left heart prosthesis could be due to acquired von Willebrand syndrome. In addition, because detection and accurate quantification of prosthetic valve dysfunction are technically challenging, abnormal results of VWF testing in a symptomatic patient with inconclusive physical findings or transthoracic echocardiography might provide an impetus for additional workup.
Accepted for Publication: January 19, 2016.
Corresponding Author: Joseph L. Blackshear, MD, Department of Cardiovascular Diseases, Mayo Clinic Florida, 4500 San Pablo Rd, Jacksonville, FL 32224 (email@example.com).
Published Online: April 6, 2016. doi:10.1001/jamacardio.2016.0075.
Author Contributions: Dr Blackshear had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Blackshear, Safford, Pollak, Stark, Rivera.
Acquisition, analysis, or interpretation of data: Blackshear, McRee, Safford, Pollak, Stark, Rivera, Wysokinska, Chen.
Drafting of the manuscript: Blackshear, Thomas, Wysokinska, Chen.
Critical revision of the manuscript for important intellectual content: Blackshear, Safford, Pollak, Stark, Rivera, Wysokinska, Chen.
Statistical analysis: Blackshear, Thomas.
Obtained funding: Blackshear.
Administrative, technical, or material support: McRee, Safford, Pollak, Stark, Thomas, Rivera, Wysokinska, Chen.
Study supervision: Blackshear.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Blackshear reported serving on an advisory board for Baxalta. No other disclosures were reported.
Funding/Support: This study was supported by grant UL1 RR024150 from the Mayo Foundation for Medical Research and Center for Translational Science Activities and by funds from the Michael S. Gordon Charitable Foundation.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentation: Dr McRee presented the abstract at ACC.15, the American College of Cardiology’s 64th Annual Scientific Session & Expo; March 16, 2015; San Diego, California.