Molecular pathogenesis and heterogeneity in type 3 VWD families in U.S. Zimmerman program

Abstract

Background:

Type 3 von Willebrand Disease (VWD) is a rare and severe form of VWD characterized by the absence of von Willebrand factor (VWF).

Objectives:

As part of the Zimmerman Program, we sought to explore the molecular pathogenesis, correlate bleeding phenotype and severity, and determine the inheritance pattern found in type 3 VWD families.

Patients/Methods:

62 index cases with a pre-existing diagnosis of type 3 VWD were analyzed. Central testing included FVIII, VWF:Ag, VWF:RCo, and VWFpp. Bleeding symptoms were quantified using the ISTH bleeding score. Genetic analysis included VWF sequencing, comparative genomic hybridization and predictive computational programs.

Results:

75% of subjects (46) had central testing confirming type 3, while 25% were re-classified as type 1-Severe or type 1C. Candidate VWF variants were found in all subjects with 93% of expected alleles identified. The majority were null alleles including frameshift, nonsense, splice site, and large deletions, while 13% were missense variants. Additional studies on 119 family members, including 69 obligate carriers, revealed a wide range of heterogeneity in VWF levels and bleeding scores, even amongst those with the same variant. Co-dominant inheritance was present in 51% of families and recessive in 21%, however 28% were ambiguous.

Conclusion:

This report represents a large cohort of VWD families in the U.S. with extensive phenotypic and genotypic data. While co-dominant inheritance was seen in approximately 50% of families, this study highlights the complexity of VWF genetics due to the heterogeneity found in both VWF levels and bleeding tendencies amongst families with type 3 VWD.

1 |. INTRODUCTION/BACKGROUND

Von Willebrand disease (VWD) is the most common inherited bleeding disorder caused by deficiency or abnormal function of von Willebrand factor (VWF) with a prevalence of approximately 1:1000.^1,2 Type 3 VWD is the rarest and most severe form of VWD characterized by undetectable levels of VWF and reduced factor VIII (FVIII).³ In contrast, type 1 VWD is the most common VWD subtype with reduced VWF levels and milder bleeding symptoms. Patients with type 3 VWD typically present with moderate to severe mucocutaneous bleeding as well as muscle hematomas and hemarthroses. While inheritance has classically been considered autosomal recessive, there is increasing evidence for co-dominant inheritance, with low VWF levels in affected heterozygous carriers.⁴ Genetic defects in type 3 VWD typically result in null alleles due to deletions, nonsense, frameshift and splice site variants located throughout the VWF gene, however missense variants have also been identified in approximately 20% of the cases.^3,4 In contrast, type 1 VWD is associated with autosomal dominant inheritance with approximately 70% due to missense variants.^5,6

The Zimmerman Program on the Biology of VWD (Zimmerman Program) is a National Institutes of Health-supported, multicenter study that enrolled VWD patients and their family members from hematology centers across the U.S. Inclusion into the study required a previous diagnosis of VWD without any other strict diagnosis parameters. As part of the Zimmerman Program study, we sought to explore the molecular pathogenesis, correlate bleeding phenotype and severity, and determine the inheritance pattern found in type 3 VWD families in the U.S.

To assess this, we examined the laboratory phenotype using multiple VWF assays and determined the bleeding phenotype using the International Society of Thrombosis and Hemostasis bleeding assessment tool (ISTH-BAT) scoring system. A thorough genetic evaluation included VWF Sanger sequencing, comparative genomic hybridization (aCGH) and in silico prediction programs. While several large type 3 studies have been previously reported,^4,7–10 the results provided here from the Zimmerman Program represents a cohort of VWD families in the U.S. with extensive phenotypic and genotypic data and provides additional insight into the bleeding tendencies, inheritance and phenotypic heterogeneity in families with type 3 VWD.

2 |. METHODS

2.1 |. Patient population

A total of 62 index cases (IC) with a pre-existing diagnosis of type 3 VWD by their local treatment center were enrolled in the Zimmerman Program through 10 primary centers and 14 secondary centers across the U.S (see Appendix). Any available family members including parents, siblings and offspring were also offered the opportunity to participate in the study. Affected family members (AFM) and unaffected family members (UFM) were assigned by the local enrolling center and had central testing to confirm their laboratory phenotype. Family members were further defined as obligate carriers (OC), who must carry a gene mutation based on their family relationship, if they either were a parent of a type 3 VWD child or were an offspring of an individual with type 3 VWD. All subjects provided informed consent and the study protocol was approved by Children’s Hospital of Wisconsin Institutional Review Board (IRB) as well as each of the local clinical acquisition center IRBs.

2.2 |. Laboratory phenotype

At the time of enrollment, whole blood was collected in 3.2% sodium citrate, processed for plasma by centrifuging twice at 2000 g for 20 min at room temperature, and stored at −80°C. All testing was performed centrally at the Versiti Blood Center of Wisconsin Hemostasis Reference Laboratory that included Factor VIII activity (FVIII), VWF antigen (VWF:Ag), VWF ristocetin cofactor (VWF:RCo), VWF propeptide (VWFpp), VWF collagen III binding (VWF:CB3), multimer analysis and reverse blood typing. Confirmatory studies were done in the research lab at Versiti Blood Research Institute (BRI) as previously described.¹¹ Phenotypic assignment was based on lab results obtained at time of enrollment. For this study, subjects were classified as low VWF (LVWF) if they had central VWF:Ag 30–50 IU/dl or VWF:RCo 30–53 IU/dl and type 1 VWD if they had VWF:Ag or VWF:RCo <30 IU/dl and a normal VWF:RCo/VWF:Ag ratio (>0.6). Type 1 VWD was further categorized into type 1 clearance (1C) for subjects with VWFpp/VWF:Ag ≥3, and type 1-severe (1S) for VWF:Ag 2–5 IU/dl, VWFpp/VWF:Ag <3.^12,13 Type 3 VWD was defined as having undetectable VWF levels (VWF:Ag and VWFpp <1 IU/dl). Subjects who had levels within the normal reference interval at time of enrollment but had a historical diagnosis of type 1 from their local treatment center, were assigned type 1H.

2.3 |. Bleeding assessment

The bleeding score (BS) was calculated using the International Society of Thrombosis and Hemostasis bleeding assessment tool (ISTH-BAT) scoring system.¹⁴ The bleeding history questionnaire was administered to each subject by a trained coordinator, nurse or physician. Subjects were encouraged but not required to answer all questions. Each bleeding symptom was scored from 0 (absence or trivial bleeding) to 4 according to the ISTH-BAT scoring key and the total BS was determined by summing the subscores.¹⁴ The established cutoffs for positive or abnormal total bleeding scores of ≥4 in adult males, ≥6 in adult females and ≥3 in children were used.¹⁵

2.4 |. VWF sequencing

EDTA whole blood was collected on each subject and DNA isolated by Qiagen Gentra Puregene method in the Molecular Diagnostic Laboratory (MDL) at Versiti. Full-length VWF Sanger sequencing was performed on all exons including intron/exon boundaries at either Harvard Partners Genome Center (HPGC), Versiti or Functional Biosciences (Madison, WI) using VWF reference sequence NM_000552.¹⁶ Raw sequence data files were processed using Softgenetics Mutation Surveyor DNA Variant Analysis Software (https://softgenetics.com/mutationSurveyor.php), and all variants were visually confirmed. If two VWF variants were not identified, cases were further investigated by aCGH at Versiti or Emory University to identify any large deletions or duplications in the VWF gene by analysis of copy number variation.¹⁷ A VWF high-resolution aCGH array was custom designed using long oligonucleotide probes approximately 60 bp in length to avoid multiple sequence variations, repeated sequences and the pseudogene. Data was manually inspected and analyzed with CytoSure Interpret Software. A multiplex PCR approach was used to map deletion breakpoints and to confirm the presence and zygosity of the deletions in the IC as well as family members. Genetic variants identified in the IC were confirmed in all the family members (including the IC) by performing direct targeted sequencing to confirm the presence of the variant.

2.5 |. In Silico analysis of variants

Putative VWF variants were analyzed by SeattleSeq (https://snp.gs.washington.edu/SeattleSeqAnnotation153/index.jsp), Fabric Genomics Opal variant interpretation software and SpliceAI. The Combined Annotation Dependent Depletion (CADD) Score, which integrates multiple annotations into one score, was used as a prediction of deleteriousness. Values ≥10 are predicted to be the 10% most deleterious substitutions, ≥20 indicate the 1% most deleterious.¹⁸ SpliceAI delta score (ranging from 0–1) was used to determine the probability of the variant being splice-altering where 0.8 is considered a high precision cutoff.¹⁹ Minor allele frequencies of variants were determined using the Genome Aggregation Database (gnomAD).²⁰

2.6 |. Statistical analysis

The statistical program GraphPad Prism (version 9.0) was used to perform all statistical analysis. Comparisons of mean laboratory values and median bleeding scores used the nonparametric Mann Whitney test and <0.05 as a cutoff for significance.

3 |. RESULTS

3.1 |. Subjects and laboratory phenotype

Of the 62 IC enrolled with type 3 VWD, 46 (75%) had central laboratory testing confirming type 3 VWD (absence of VWF), six subjects (10%) had detectable VWF:Ag and VWFpp levels between 2–5 IU/dl and were re-classified as type 1S, nine subjects (15%) had a significantly increased VWFpp/VWF:Ag and were re-classified as type 1C and one subject had variable levels with unclassified VWD. Two of the 46 confirmed type 3 IC did not have DNA available, therefore 44 type 3 IC were further evaluated by genotypic analysis (Figure 1). This group was comprised of an equal number (22) of females and males with a mean age at enrollment of 24 (range 0–61). The majority were Caucasian (82%).

FIGURE 1.

von Willebrand factor type 3 subjects enrolled in the Zimmerman Program. Sixty-two subjects with a pre-existing diagnosis of type 3 VWD were enrolled and evaluated in the Zimmerman Program. Nine subjects (15%) had a significantly increased VWFpp/VWF:Ag and were re-classified as type 1 clearance (1C) and six subjects (10%) had detectable VWF:Ag and VWFpp levels between 2–5 IU/dl and were re-classified as type 1-Severe (1S). 46 (75%) of the subjects had central laboratory testing confirming type 3 (absence of VWF). Two of the 46 confirmed type 3 cases did not have DNA available. Therefore 44 subjects were evaluated by genotypic analysis; 39 subjects had additional family members available to study

3.2 |. Genotype analysis

Genotype was determined for 44 subjects by examining results from Sanger VWF sequencing and aCGH where candidate VWF variants were included if present in <1% of Zimmerman Program healthy controls.¹⁶ VWF variants were identified in all of these subjects: 36 subjects had two variants (82%), six subjects had only one (14%), and two had >2 (5%); 82 of the 88 allele variants expected were identified (93%). Homozygous variants were found in seven of the subjects who did not have an additional large VWF deletion. Figure 2 shows the location of the 60 different variants that are spread throughout the gene, including 34 candidate variants not previously reported as well as recurrent variants. Of the genetic variants identified, 28% were frameshift, 27% nonsense, 17% splice site, 13% missense, 9% large deletions detected by aCGH, and 6% intronic. The large deletions included heterozygous deletions of exon 1–3, exon 4–5, exon 18 and exon 35–38 in combination with another VWF variant. One case had different deletions (exon 1–3 and exon 4–5) on separate alleles. The breakpoints of the deletions were resolved for exon 1–3 deletion (c.−30029_220+3487del), exon 4–5 (c.221–977_532+7059del), exon 18 (c.2282−809_2442+2811delinsT) and exon 35–38 deletion (c.5843−2754_6799–1517del).

FIGURE 2.

Distribution of the 60 different genetic variants identified in the Zimmerman Program type 3 VWD subjects by VWF domain. Variants included deletions, nonsense, splice site, missense and intronic changes throughout the gene. Candidate variants not previously reported are in bold and recurrent variants are underlined. (VWF domain structure credit Springer 2014³⁷)

The in silico analysis of the VWF variants is shown in Table 1. All putative variants identified were rare with frequencies from 0.0004%–0.26% or not present (NA) in gnomAD. CADD Scores were evaluated as a predictor of deleteriousness. The majority (70%) of the variants were nonsense, frameshift and large deletions that result in null alleles and are presumed to be deleterious. Of the 12 missense variants, 7 (p. Phe165Val, p. Tyr301Cys, p. Ala634Pro, p. Cys851Tyr, p. Asp896Tyr, p. Ile1343Val and p. Gly1828Val) were predicted to be deleterious with a score ≥20 and the other 5 had a score >14 (p. Cys295Tyr, p. Met814Ile, p. Trp1144Gly, p. Gln1346Arg and p. Cys2739Arg). SpliceAI was used to assess the probability of the variant being splice-altering.¹⁹ 7 of the splice site region variants (c.3675−1G>A, c.5621−2A>C, c.6256+2dup, c.6598+1G>A, c.6798+1G>A, c.7730−2del and c.7770+1G>A) and one intron variant (c.324−10_324−9insCAGAGT) were predicted to alter the splice site either due to splice-acceptor or splice-donor loss with scores ≥0.9. One missense variant p. Gly1828Val was also predicted to be a gain of a splice-acceptor site.

TABLE 1.

In silico analysis of type 3 variants

rsID	HGVS c	HGVS p	Consequence	CADD	gnomAD	SpliceAl
rs61753984	c.lOOC>T	p. Arg34*	STOP GAINED	38.00	1.2E-05
-	c.324−14_324−4delinsAAGTTCAGAGTCT		INTRON VARIANT	NA	NA	NA
-	c.324−10_324−9insCAGAGT		INTRON VARIANT	NA	NA	0.98 (Acceptor loss)
rs754520488	c.493T>G	p. Phe165Val	MISSENSE VARIANT	22.60	4.0E-06
rs748853071	c.670C>T	p. Gln224*	STOP GAINED	39.00	NA
rs773976969	c.875−3C>A		SPLICE REGION VARIANT	7.40	4.0E-06	0.52 (Acceptor loss)
rs770781601	c.884G>A	p. Cys295Tyr	MISSENSE VARIANT	16.49	4.0E-06
-	c.902A>G	p. Tyr301Cys	MISSENSE VARIANT	26.6	6.6E-06	0.07 (Acceptor loss)
rs147924974	c.992_993delinsAA	p. Cys331*	STOP GAINED	40.00	2.6E-05
rs62643625	c.H17C>T	p. Arg373*	STOP GAINED	39.00	6.6E-06	0.01 (Acceptor gain)
-	c.1352_1373del	p.L451Rfs*19	FRAM ESHIFT VARIANT	20.00	NA
-	c.1682_1722dup	p. Arg575Thrfs*16	FRAMESHIFT VARIANT	20.00	NA
rs760570393	c.1730−10C>A		INTRON VARIANT	17.7	5.3E-05	0.75 (Acceptor loss)
-	c.1900G>C	p. Ala634Pro	MISSENSE VARIANT	24.50	6.6E-06	0.01 (Donor loss)
-	c.1993del	p. Cys665Alafs*3	FRAM ESHIFT VARIANT	20.00	NA
rs764755360	c.2067C>A	p. Cys689*	STOP GAINED	40.00	8.0E-06
rs267607309	c.2072del	p. Pro691Glnfs*50	FRAM ESHIFT VARIANT	23.70	6.6E-06
rs61748465	c.2269_2270del	p. Leu757Valfs*22		28.00	6.6E-06
rs62643632	c.2435del	p. Pro812Argfs*31	FRAM ESHIFT VARIANT	25.50	5.9E-05	0.02 (Donor loss)
rs758895722	c.2438dup	p. Met814Hisfs*5	FRAM ESH IFT VARIANT	26.90	2.0E-05	0.03 (Donor loss)
-	c.2442G>C	p. Met814lle	MISSENSE VARIANT	15.71	NA
-	c.2552G>A	p. Cys851Tyr	MISSENSE VARIANT	24.60	NA
rs1022033438	c.2686G>T	p. Asp896Tyr	MISSENSE VARIANT	32.00	6.6E-06	0.03 (Acceptor loss)
-	c.2908del	p. Leu970Serfs*9	FRAMESHIFT VARIANT	20.00	NA
rs61748495	c.3108+5G>A		SPLICE REGION VARIANT	21.90	1.3E-05	0.59 (Donor gain)
-	c.3337C>T	p.Gln1113*	STOP GAINED	49.00	6.6E-06	0.01 (Donor loss)
-	c.3346A>T	p. Lys1116*	STOP GAINED	42.00	NA
rs267607325	c.3430T>G	p. Trp1144Gly	MISSENSE VARIANT	18.24	NA
rs746457842	c.3675−1G>A		SPLICE ACCEPTOR VARIANT	15.25	4.2E-06	0.99 (Acceptor loss)
rs267607337	c.3931C>T	p.Gln1311*	STOP GAINED	42.00	1.3E-05
rs150923481	c.4027A>G	p. lle1343Val	MISSENSE VARIANT	20.00	1.4E-04
rs778358708	c.4037A>G	p. Gln1346Arg	MISSENSE VARIANT	14.88	1.2E-05
rs267607345	c.4636del	p. Val1546*	FRAMESHIFT VARIANT	20.00	8.0E-06
rs61750112	c.4696C>T	p. Arg1566*	STOP GAINED	45.00	8.0E-06
rs61750595	c.4975C>T	p. Arg1659*	STOP GAINED	38.00	2.0E-05
-	c.5128_5132del	p. Met1710Glufs*31	FRAMESHIFT VARIANT	20.00	NA
-	c.5189dup	p. Ser1731Valfs*12	FRAMESHIFT VARIANT	20.00	NA
rs1299445904	c.5483G>T	p. Gly1828Val	MISSENSE VARIANT	33.00	6.6E-06	0.97 (Acceptor gain)
rs61750612	c.5557C>T	p. Arg1853*	STOP GAINED	42.00	1.3E-05
rs769632868	c.5621−2A>C		SPLICE ACCEPTOR VARIANT	13.65	4.0E-06	0.98 (Acceptor loss)
-	c.5942dup	p. Gln1982Alafs*9	FRAM ESH IFT VARIANT	20.00	NA
-	c.6043G>T	p. Glu2015*	STOP GAINED	44.00	NA
-	c.6256+2dup		SPLICE REGION VARIANT	NA	NA	0.96 (Donor Loss)
-	c.6598+1G>A		SPLICE DONOR VARIANT	16.87	NA	0.99 (Donor Loss)
-	c.6652del	p. Arg2218Glyfs*7	FRAM ESH IFT VARIANT	20.00	NA
-	c.6798+1G>A		SPLICE DONOR VARIANT	24.50	NA	0.99 (Donor Loss)
rs202001513	c.6901+14C>T		INTRON VARIANT	8.65	2.6E-03	0.0100 (Donor loss)
-	c.7274dup	p. Cys2425Trpfs*50	FRAMESHIFT VARIANT	20.00	NA
-	c.7448dup	p. Tyr2483*	STOP GAINED	20.00	NA
rs61751296	c.7603C>T	p. Arg2535*	STOP GAINED	48.00	1.97E-05
rs1368264172	c.7627C>T	p. Gln2543*	STOP GAINED	47.00	4.0E-06	0.0600 (Acceptor gain)
rs1055962685	c.7730−2del		SPLICE ACCEPTOR VARIANT	26.3	6.6E-06	0.950 (Acceptor loss)
rs372396117	c.7730−3C>G		SPLICE REGION VARIANT	22.90	1.3E-05	0.800 (Acceptor loss)
rs200770256	c.7770+1G>A		SPLICE DONOR VARIANT	33.00	6.6E-06	0.920 (Donor loss)
-	c.8215T>C	p. Cys2739Arg	MISSENSE VARIANT	14.16	NA
rs1414503929	c.8419_8422dup	p. Pro2808Leufs*24	FRAMESHIFT VARIANT	33.00	6.6E-06

Abbreviations: CADD, Combined Annotation Dependent Depletion Score; gnomAD, Genome Aggregation Database minor allele frequency; HGVS c, Human Genome Variation Society coding DNA reference sequence; HGVS p, HGVS protein reference sequence; rsID, Reference SNP cluster ID; SpliceAI, splice delta score.

3.3 |. Bleeding evaluation

Individual bleeding scores as well as laboratory and genetic findings for all the type 3 ICs are shown in Table 2. Figure 3 illustrates that type 3 subjects had an abnormal BS according to age and gender published cutoffs, which is ≥6 in adult females, ≥4 in adult males and ≥3 in pediatric subjects <18 years of age.¹⁵ The median BS in the type 3 cohort was 15, with no significant difference between females (16) and males (15). While adult females (n = 14) had an increased score of 24 compared to adult males (n = 12) BS of 17, it was not statistically significant (p = .1196). Pediatric (<18) females (n = 7) and males (n = 10) had a similar BS of 11 and 11.5 respectively. There was a difference in total BS between adult females and pediatric females (p < .01) and adult males and pediatric males (p < .05).

TABLE 2.

Phenotypic and genotypic data of the 44 type 3 index cases

Study ID	Age/Gender	FVIII	VWF:Ag	ABO	ISTH BS	Pos/Neg BS	VWF Variant 1	VWF Variant 2
ZA0026	34/F	4	<1	O	15	+	p. Glu2015*	c.7730−2del
ZA0216	58/F	7	<1	A	31	+	c.221−977_532+7059del	NSV
ZA0267	23/F	4	<1	A	24	+	p. Arg1566*	c.1730−10C>A
ZB0016	47/F	3	<1	A	21	+	p. Gln2543*	p. Pro812Argfs*31
ZE0005	8/F	3	<1	B	5	+	p. Arg373*	C.3675−1G>A
ZF0017	21/M	8	<1	O	15	+	p. Gln1982Alafs*9	c.2282−809_2442+2811delinsT
ZF0092	4/M	6	<1	O	6	+	p. Cys851Tyr	NSV
ZF0145	21/F	a	a	B	28	+	p. Tyr301Cys	p. Tyr301Cys
ZF0164	16/F	a	a	A	11	+	p. Phe165Val	p. Lys1116*
ZF0171	8/F	2	<1	NA	4	+	c.6798+1G>A	C.6798+1G>A
ZG0002	18/F	2	<1	A	9	+	c.6901+14C>T (homozygous)	c.7730−2del (homozygous)
ZI0002	61/F	a	a	A	28	+	c.5621−2A>C	p. Arg2535*
ZI0003	45/F	a	a	O	24	+	c,875−3C>A	p. Cys295Tyr
ZJ0005	11/M	3	<1	O	18	+	c.7770+1G>A	C.7770+1G>A
ZL0020	48/M	8	<1	O	10	+	p. Cys331*	c.324−14_324−4delinsAAGTTCAGAGTCT
ZL0079	20/M	4	<1	O	21	+	p. Pro691Glnfs*50	c.2282−809_2442+2811delinsT
ZL0146	1/F	0	<1	O	6	+	p. Arg575Thrfs*16	p. Val1546*
ZM0034	24/M	NA	<1	A	22	+	p. Met814Hisfs*5	c.324−14_324−4delinsAAGTTCAGAGTCT
ZM0209	44/F	6	<1	A	29	+	p. Met814lle	p. Tyr2483*
ZM0242	47/M	7	<1	A	18	+	c.3108+5G>A	c.3108+5G>A
ZM0393	41/F	NA	<1	A	24	+	p. Met814Hisfs*5	p. Arg1566*
ZM0499	12/M	3	<1	O	15	+	p. Pro2808Leufs*24	NSV
ZM0515	2/M	4	<1	NA	10	+	p. Pro812Argfs*31	C.6598+1G>A
ZM0852	24/M	2	<1	B	11	+	c.324−10 324−9insCAGAGT	p. Arg2535*
ZN0043	11/F	3	<1	NA	13	+	p. Arg34*	p. Asp896Tyr
ZN0087	25/F	9	<1	O	29	+	p. Leu757Valfs*22	NSV
ZN0088	40/F	5	<1	O	16	+	p. Arg575Thrfs*16	c.221−977_532+7059del
ZN0101	38/M	6	<1	O	16	+	p. Cys689*	p. Pro812Argfs*31
ZN0646	17/F	2	<1	B	NA	NA	p. Leu451Argfs’19	p. Arg1659*
Z00056	14/F	4	<1	NA	19	+	p. Ser1731Valfs*12	p. Arg1853*
Z00068	18/M	NA	<1	O	9	+	p.Gln224*	p. Arg1659*
ZPOOOl	1/M	8	<1	AB	8	+	p. Gly1828Val	c.7730−3C>G
ZQ0004	27/F	3	<1	B	6	+	p. Leu970Serfs*9	p. Leu970Serfs*9
ZROOOl	O/M	2	<1	NA	4	+	c.6256+2dup	p. Trp1144Gly
ZR0008	34/F	5	<1	O	10	+	p. Cys665Alafs*3	p. Gln1113*
ZS0020	5/F	a	a	NA	11	+	p. Met814Hisfs*5	p. Arg1659*
ZTOOOl	40/M	4	<1	O	24	+	p. Arg1853*	p. Gln1311*
ZU0021	35/M	8	<1	O	23	+	p. Arg2535*	p. Arg2535*
ZU0081	24/M	2	<1	A	18	+	c.221−977_532+7059del	c.−30029_220+3487del
ZU0092	5/M	NA	<1	NA	8	+	p. Met814Hisfs*5	c.5843−2754_6799~1517del
ZV0007	15/M	a	a	O	13	+	p. Arg2218Glyfs*7	p. Cys2739Arg
ZW0012	7/M	a	a	O	13	+	C.3675−1G>A	NSV
ZX0008	46/M	a	a	O	15	+	p. Cys2425Trpfs*50	NSV
ZY0006	5/M	3	<1	O	20	+	p. Ala634Pro	p. Met1710Glufs*31

Note: FVIII activity (FVIII), VWF antigen (VWF:Ag), and blood type (ABO) were measured centrally at time of enrollment. VWF ristocetin cofactor activity (VWF:RCo) was <10 in all untreated individuals.

ISTH BS, ISTH BAT bleeding score.

Positive (pos, +) or Negative (Neg) bleeding score determined by age and gender cutoffs.¹⁵

First sequence variant (VWF Variant 1) and second variant (VWF Variant 2) identified in each subject are listed, or NSV for no causative sequence variant found in VWF.

Abbreviations: F, Female; M, Male; NA, lab values not available.

^aTreated sample.

FIGURE 3.

Abnormal bleeding score in type 3 subjects. Forty-three type 3 subjects reported an abnormal bleeding score (ISTH BS) according to age and gender published cutoffs¹⁵ (one subject was unreported). There was no significant difference seen in total BS between adult females and males, however there was a difference amongst adult and pediatric (Ped <18 years of age) subjects

The ISTH BS consists of 14 subcategories of bleeding: epistaxis, cutaneous, minor wound, oral cavity, tooth extraction, gastrointestinal (GI) bleeding, hematuria, surgical, menorrhagia, postpartum, central nervous system (CNS), hemarthrosis, hematoma and other. Subscores ranged from 0–4 based upon bleeding episodes and medical attention or intervention.¹⁴ In our analysis, the severity of bleeding symptoms was determined by evaluating each individual bleeding category using subscores >1. The most frequently reported bleeding symptoms were due to epistaxis, cutaneous, oral cavity, and hemarthrosis as well as menorrhagia in women (Table 3). Similar bleeding symptoms were reported in males and females, however, males reported more hemarthrosis and hematomas, 83% and 67%, respectively, while 70% of females reported menorrhagia and 64% surgical bleeding.

TABLE 3.

Frequency and severity of bleeding symptoms in type 3 subjects

Note: Frequency of bleeding represented by darker shades ranging from none (0%) to darkest shade (>70%) in type 3 VWD subjects by age (adult and pediatric<18 years of age) and gender (female [F], and male [M]).

Abbreviations: CNS, Central Nervous System; GI, Gastrointestinal.

Comparison of the ISTH-BAT BS for subjects by VWF variant showed no difference in BS according to number or location of the variant. There was a slight difference (p < .05) between the 33 subjects who had a null allele and bleeding score of 16 compared to those without null alleles and bleeding score of 10, however that could be due to the age difference in these groups (24 and 9.5 respectively). There was also no difference in ISTH BS of IC with variants in the propeptide region (16) compared to other regions of VWF (15).

3.4 |. Family studies

Additional family members in 39 of the unrelated index cases were further studied. The family member cohort consisted of 77 affected family members (AFM) and 42 unaffected family members (UFM). The relationships to the index included parents (47%), siblings (25%), offspring (9%), grandparents (5%) and other extended family (13%). All family members had central laboratory testing performed and a bleeding assessment at time of enrollment (Table S1). A summary of the demographic characteristics, clinical phenotyping and study entry ISTH-BAT BS are detailed in Table 4. The AFM cohort consisted of 64% low VWF, 21% type 1, 10% type 3, 4% type 1H and 1% type 1C. The median BS in AFM was 1 (range 0–46), which was not significantly different from BS of 1 (range 0–9) in UFM. Type 3 AFM (n = 8) from seven different families had a median BS of 22.5 that was not statistically different (p = .1164) from BS in the type 3 ICs.

TABLE 4.

Summary of subject demographic and clinical characteristics

	AFM	UFM	OC
No. Subjects	77	42	68
No. Females (%)	42 (54)	27 (64)	37 (54)
Mean Age (range)	33 (0–75)	44 (3–87)	43 (0–81)
No. Pediatric <18 (%)	20 (26)	5 (12)	6 (9)
Median bleeding score (range)	1 (0–46)	1 (0–9)	1 (0–11)
No. Blood Group O (%)a	44 (59%)	9 (20%)	30 (44%)
Mean FVIII (range)b	56 (2–149)	122 (71–200)	86 (21–174)
Mean VWF:Ag (range)b	27 (<1–81)	89 (56–211)	48 (12–155)
Mean VWF:RCo (range)b	31 (<1–83)	88 (57–201)	47 (13–160)

Note: Central Factor VIII activity (FVIII), VWF antigen (VWF:Ag), VWF ristocetin cofactor activity (VWF:RCo) and bleeding score obtained at study entry.

Abbreviations: AFM, affected family members; OC, obligate carriers; UFM, unaffected family members (UFM).

^aSubjects who were too young or unable to obtain blood type were excluded.

^bEight type 3 treated subjects were excluded from the mean and range determinations.

Genetic variants identified in the IC were assessed in all the family members by performing direct targeted sequencing to determine the presence or absence of the variant and to surmise the inheritance pattern. There were 69 obligate carriers (OC), who were either a parent with a type 3 VWD child or an offspring of a parent with type 3 VWD, where we confirmed that 64 (93%) did carry one of the putative variants identified in the IC. In four of the families where only one variant was identified, the no sequence variant (NSV) allele was linked to three OCs with low VWF/type 1 phenotype and two OCs with a normal phenotype. No de novo variants were identified. Affected OC (n = 45) had mean VWF:Ag and VWF:RCo of 37, which was significantly different (p < .0001) from unaffected OC (n=24) with mean VWF:Ag of 78 and VWF:RCo of 76. Co-dominant inheritance was present in 51% of families and recessive in 21%, however 28% were ambiguous and in some cases had evidence of both types of inheritance.

Seven of the type 3 ICs had homozygous VWF variants (p. Tyr301Cys, c.6798+1G>A, c.7730−2del, c.7770+1G>A, c.3108+5G>A, p. Leu970Serfs*9 and p. Arg2535*) where we were able to confirm that the same variant was inherited from both parents on separate alleles in four families; in the other three cases we only had one parent or neither parent available to study. Of the 60 different variants identified, nine of these were null alleles that were found in multiple families (p. Arg575Thrfs*16, exon 4–5 deletion, exon 18 deletion, p. Pro812Argfs*31, p. Met814Hisfs*5, p. Arg1659*, p. Arg1853*, p. Arg2535*, c.7730-2del). Obligate carriers with these variants displayed a range of expressivity with VWF levels ranging from normal to type 1 VWD as well as varied bleeding scores (Table 5).

TABLE 5.

Null variants in multiple families displaying variable phenotypic heterogeneity

SV	Family #	Phen Dx	Age/Gender	VWF:Ag	VWF:Rco	ABO	ISTH BS
p. Arg575Thrfs*16	ZL 046	Low VWF	29/M	36	33	O	0
	ZN 009	Type 1H	61/F	81	83	A	11
Exon 4–5 Deletion	ZU 029	Normal	62/M	99	83	A	0
	ZA 048	Normal	31/M	59	58	A	0
	ZN 009	Type 1	64/M	30	NA	NA	0
Exon 18 Deletion	ZF 003	Normal	62/M	73	57	O	0
	ZL 025	Normal	44/F	67	74	O	0
p. Pro812Argfs*31	ZM 086	Normal	23/F	56	63	O	0
	ZN 018	Low VWF	67/M	42	45	O	3
p. Met814Hisfs*5	ZS 005	Low VWF	33/F	61	51	A	4
	ZU 034	Type 1	49/M	28	37	O	0
	ZM 002	Type 1H	44/M	73	77	A	1
	ZM 068	Low VWF	59/F	54	40	A	6
p. Arg1659*	ZO 012	Low VWF	48/M	43	45	O	0
	ZN 138	Low VWF	51/F	44	34	O	0
	ZS 005	Low VWF	37/M	59	52	NA	0
p. Arg1853*	ZT 001	Normal	68/M	91	77	B	0
	ZO 009	Type 1	42/F	29	30	A	8
	ZO 009	Low VWF	64/M	32	42	O	1
p. Arg2535*	ZU 001	Type 1	3/M	16	21	O	6
	ZU 001	Type 1	5/M	21	27	O	5
	ZU 001	Low VWF	16/M	51	52	O	7
	ZI 002	Normal	65/F	62	63	A	4
c.7730-2del	ZA 003	Low VWF	38/U	56	39	AB	6
	ZG 004	Normal	33/F	100	95	A	1

Note: VWF antigen (VWF:Ag), VWF ristocetin cofactor activity (VWF:RCo), blood type (ABO) and bleeding score (ISTH BS) obtained at time of study entry.

Abbreviations: Age, age at enrollment; F, Female; M, Male; Phen Dx, Phenotypic diagnosis; SV, VWF Sequence Variant; U, Unknown; NA, lab values not available.

4 |. DISCUSSION

While several type 3 studies have been previously reported,^4,7–10 the results provided here from the Zimmerman Program on 44 type 3 index cases and 119 family members, represents one of the first large cohorts from the U.S. describing VWD families with extensive phenotypic and genotypic data. Our findings revealed that while the majority of candidate VWF variants resulted in null alleles, obligate carriers displayed a wide range of heterogeneity in VWF levels and bleeding scores, even amongst those with the same variant. This report also highlights the complex inheritance patterns in VWD with approximately 50% of the families displaying co-dominant inheritance.

Although inclusion criteria required a preexisting diagnosis of VWD, all subjects had thorough central laboratory testing performed to determine their phenotypic assignment based upon analysis of all available data, including historic levels. We found that 25% of this cohort showed evidence of type 1-severe or type 1C but were diagnosed as type 3 (Figure 1, Table S2). In the six type 1-severe cases, all had detectable VWF:Ag levels between 2–5 IU/dl with a parallel reduction of VWFpp level and two of these subjects had follow-up laboratory evaluations available with consistent VWF:Ag levels in this range. The nine type 1C cases had reduced VWF:Ag and increased VWFpp/VWF:Ag >3, without indication of treatment. The ability to distinguish type 1C from type 3 is important as these subjects should have normal Weibel-Palade bodies and α-granule stores of VWF, and a milder bleeding phenotype where DDAVP could be useful for minor bleeding episodes.²¹ There are clinical implications in differentiating type 3 patients from type 1-severe who produce VWF, in the approach to treatment²² and the unlikely chance of alloantibody inhibitor formation. These type 1C and 1-severe cases are being further analyzed and are part of ongoing studies in the Zimmerman Program.

A possible explanation for the discrepancy between preexisting diagnosis and phenotypic assignment at enrollment could be the lack of availability and use of the VWFpp assay. This was also seen In the WIN study, where authors found that 41% of their type 3 cases had detectable VWFpp levels and evidence of type 1-severe or type 1C.²² Another factor is the sensitivity of the VWF assays done historically, which could make it difficult to distinguish between severely reduced but detectable VWF levels, as opposed to the absence of VWF to define type 3 VWD. We do have cases where the subject was drawn while being treated with factor replacement. In the majority of these cases, we were able to use untreated follow-up samples to confirm laboratory values, however in seven cases we had to rely on historical data collected by the local site, which had variable lower limits of detection. In one case the discrepancies in the historical lab values resulted in the designation of unclassified VWD.

Causative VWF variants were identified in all 44 type 3 subjects with 93% of the mutant alleles accounted for, which is consistent with other type 3 VWD studies that have reported between 80%–95% of cases with pathogenic variants.^4,7–10 Of the variants identified, 9% were large heterozygous deletions of exon 1–3, exon 4–5, exon 18 and exon 35–38. The exon 4–5 in-frame deletion has been previously reported as the most common deletion in the VWF gene occurring in both type 1 and type 3 VWD.²³ This deletion was also the most common in our cohort having been found in three ICs as well as two type 3 AFM, one type 1 OC and two OC with normal VWF levels and no bleeding. An exon 1–3 deletion has been reported as the most common cause of type 3 VWD in Hungary.²⁴ We confirmed an exon 1–3 deletion in one IC, however our breakpoints (c.−30029_220+3487del) were different from the VWF database (VWFdb: EAHAD Coagulation Factor Variant Databases [EAHAD-CFDB] https://vwf-db.eahad.org/) published breakpoints (c.−30051_220+3465del).²⁵ Interestingly, the exon 1–3 deletion was found in addition to the exon 4–5 deletion inherited on different alleles in a recessive manner. The exon 18 deletion was previously reported in the VWFdb by groups in Italy and Poland, however no breakpoints were reported. We resolved the breakpoints in the two ICs, c.2282−809_2442+2811delinsT, and confirmed the out-of-frame deletion in two normal OCs. The other partial gene deletion was a novel exon 35–38 out-of-frame deletion in which we determined the breakpoints and confirmed the presence of the deletion in one normal OC and two AFM with type 1/low VWF.

Several of the null variants have been previously reported as frequently occurring in other populations. The variant p. Arg2535* was found in two (5%) unrelated families and has previously been reported in patients of Dutch, German, Italian, Swedish and Turkish ancestry.^3,10 The frameshift variant identified in the original VWD family from the Aland Islands, c.2435del (p. Pro812Argfs*31), and recently described as the most common type 3 variant in Northern Europe in 3WINTERS-IPS, was only found in two families along with another variant.^10,26 The variant p. Pro2808Leufs*24, observed in the Canadian type 3 VWD study and associated with milder bleeding, was found in one family, however our IC presented with a more severe bleeding phenotype (ISTH BS 15).²⁷ While some of the variants identified were previously reported and frequently occurring in other populations, we also identified 34 novel variants (Figure 2).

The homozygous variants c.7730−2del, c.3108+5G>A, c.7770+1G>A, and p. Leu970Serfs*9, were found, respectively, in families of African American, Caucasian and two of Asian descent. The prevalence of homozygous variants in U.S. type 3 subjects (16%) was lower than the occurrence observed in European (50%) and Iranian (84%) type 3 VWD, which may reflect a higher rate of consanguinity in those populations.¹⁰

Nine of the genetic variants were found in multiple families where OC displayed heterogeneity from normal to low VWF to type 1 as well as varied bleeding scores (Table 5). This exemplifies the complexity when correlating the genotype and phenotype and suggests that in addition to the unaffected allele, other factors including age, blood group, carbohydrate modifications and modifying genes contribute to VWF levels and bleeding.^28–31

Missense variants accounted for 13% of the type 3 VWF variants identified, which is slightly lower than those found in the Canadian (19%) and 3WINTERS-IPS (21%) studies.^4,10 The underlying mechanisms of these missense variants are suspected to interfere with VWF synthesis, secretion, or clearance. One index case had a missense variant associated with increased clearance of VWF, p. Trp1144Gly, along with a splice site region variant (c.6256+2dup) that resulted in absence of VWF.¹² The mother with p. Trp1144Gly exhibited the classical type 1C phenotype of increased VWFpp/VWF:Ag while the two family members with the splice region variant had low to normal VWF levels. This suggests that while the type 3 index case appears to have two loss-of-function alleles, the p. Trp1144Gly allele needs to hetero-multimerize with a normal allele to be secreted.

While type 3 VWD is typically considered a recessive disease, we found that some families exhibited co-dominant inheritance with OC having both low VWF levels and bleeding. This is demonstrated by a family where the IC inherited p. Ser1731Valfs*12 from a father with low VWF (VWF:Ag 44, ISTH BS 6) and p. Arg1853* from the mother with type 1 VWD (VWF:Ag 29, ISTH BS 8). In our study, we found that more than half of the OC were diagnosed with type 1 VWD or low VWF, which is consistent with the 48% of OC with VWD in the Canadian type 3 study.⁴ More nonsense and frameshift variants (81%) were observed in families who appear to have co-dominant inheritance patterns, whereas those with recessive or mixed inheritance have more splice, missense and large deletion variants. Families with apparent mixed inheritance could be due to variable expressivity, contribution of the unaffected allele, or an age effect with levels of VWF increasing over time.^32–34 This is demonstrated in a family where the p. Val1546* variant is found in a 55 year old maternal grandmother with normal levels (VWF:Ag 56), a 27 year old mother with low VWF (VWF:Ag 38) and a 3 year old with type 1 VWD (VWF:Ag 30). However, this is not always the case, and we observe no age affect in levels in other families.

Two of the type 3 cases had splice site variants that were only identified upon re-analysis of the Sanger sequencing results. In one case, the homozygous splice site variant c.7770+1G>A was missed by analysis software but confirmed after manual review of the Sanger data files. This variant was subsequently sequenced and verified in both the IC and family members and has been reported associated with type 3 where multiple splicing events lead to premature stops in VWF.³⁵ A second case was identified through analysis software, but never flagged as a potential causative variant during analysis. This variant, c.3675−1G>A, was also confirmed in original sequencing file and subsequently confirmed in the mother of the IC.

One limitation to our study is that we did not perform full VWF exonic sequencing on family members, therefore it is possible that we may have missed another variant that could affect VWF levels and bleeding. There were also a small number of missing variants in this cohort, which may be due to the fact that only the coding regions and intron/exon boundaries were sequenced. Further investigation of these cases may identify variants in the promoter, deep intronic or other modifying genes. It is also possible that a combination of rare and common VWF variants or variants of uncertain significance could contribute to the VWD phenotype. These studies are ongoing in the Zimmerman Program to further analyze type 3 patients with only 1 VWF variant by whole genome sequencing (WGS), which will shed light on additional pathogenic variants or genetic modifiers affecting VWF levels.

Median BS in the type 3 cohort was 15, which is consistent with those reported from other type 3 studies even when different BATs were used.^4,24,36 The difference seen between adult and pediatric females (Figure 3) was not only due to menorrhagia and postpartum bleeding as evidenced by a median score of 19.5 (p < .05) when excluding those categories in adults. Differences can be seen in female adults reporting more bleeding from surgery, hemarthrosis, oral cavity, minor wound and epistaxis compared to pediatric females (Table 3).

The most severe and frequently reported bleeding symptoms were due to epistaxis, cutaneous, oral cavity bleeding, hemarthrosis, and menorrhagia in women, which is similar to other type 3 studies.^8,36 While similar bleeding symptoms were reported in males and females, there were differences in males reporting more hemarthrosis and hematomas, 83% and 67%, respectively, while 79% of females reported menorrhagia and 64% surgical bleeding. In the females who reported surgical bleeding, 50% of bleeding episodes were due to gynecological, abdominal, or orthopedic surgery while 50% were a mix of other types of surgical procedures. There were also differences seen in most of the categories between the adult and pediatric populations by gender. While children reported similar cutaneous bleeding and hemarthrosis, there were differences seen with boys experiencing more epistaxis, oral cavity, and minor wound bleeding compared to girls (Table 3).

There was no difference in BS between AFM vs UFM, which was similar to findings in the Canadian study with OC BS of 1 compared to UFM score of 0⁴ This could be influenced by the number of pediatric subjects in these cohorts, 26% vs 12%, respectively. While it might seem surprising for AFM to have low VWF levels but normal BS, we have previously reported that 24% of the Zimmerman Program type 1 cohort with VWF levels <30 IU/dl had an ISTH BS in the normal range.¹¹ More of the AFM did have blood group O (59%) compared to the UFM (20%), which could contribute to their lower levels. It is also possible that some of the AFM were diagnosed based on a family member with VWD, a family history of bleeding, and low VWF levels, but may not have had a personal history of bleeding or enough bleeding challenges to contribute to an abnormal BS.

A potential limitation is that the bleeding score includes the entire lifetime of the subject’s bleeding history. Since these subjects were enrolled with a pre-existing diagnosis, their BS could have been influenced by their age at diagnosis, treatment and number of hemostatic challenges, and recall bias. In addition, parents who provide the bleeding history for their children may be more comprehensive than those of adults who need to recall all bleeding episodes.

5 |. CONCLUSIONS

While a number of large type 3 VWD studies have been previously reported, the results provided here from the Zimmerman Program represents one of the first cohorts in the U.S. with extensive phenotypic and genotypic data and provides additional insight into the bleeding tendencies, inheritance, and heterogeneity in families with type 3 VWD. We found that some type 1-severe and type 1C patients have been diagnosed as type 3, which may have clinical and treatment implications. We identified 60 different candidate VWF variants in this cohort where the majority resulted in null alleles, however 7% of type 3 single alleles did not have an attributable variant within the VWF locus. Bleeding symptoms are severe in type 3 and are predominantly due to epistaxis, cutaneous, oral cavity bleeding, hemarthrosis and menorrhagia, however we have observed variable bleeding severity and symptoms by gender and age. Co-dominant inheritance was present in approximately 50% of the families and null variants were common in these families, however the genetics is complicated by the heterogeneity found in both VWF levels and bleeding within the same family and amongst individuals from different families with the same variant. Continued prospective studies of these patients will shed light on other genetic variants or modifying genes affecting VWF levels and provide valuable insight into the variability and changes in bleeding in type 3 VWD families over time.

ESSENTIALS.

• Type 3 VWD is a rare and severe form of VWD characterized by undetectable levels of VWF

• Both phenotype and genotype were investigated in type 3 VWD families in US Zimmerman Program

• The majority of VWF variants caused null alleles but carriers had variable levels and bleeding

• This study provides insight into molecular pathogenesis and bleeding tendencies in type 3 VWD

APPENDIX

ZIMMERMAN PROGRAM INVESTIGATORS

Principal Investigators include:

R. Montgomery, V. Flood, S. Haberichter, TAbshire, HWeiler, Versiti Blood Research Institute, Milwaukee, WI; DLillicrap, PJames, Queen’s University, Kingston, ON, Canada; JO’Donnell, Royal College of Surgeons in Ireland, Dublin, Ireland, CNg, University of Colorado, Denver, CO; J. Di Paola, B. Sadler, Washington University in St. Louis, St. Louis, MO.

Directors of the primary centers include:

T. Abshire, CBennett, RSidonio, Emory University School of Medicine, Atlanta, GA; MManco-Johnson, J. Di Paola, C. Ng, Mountain States Regional Hemophilia and Thrombosis Center, Aurora, CO; JJourneycake, AZia, UT Southwestern, Dallas, TX; JLusher, MRajpurkar, Wayne State University, Detroit, MI; AShapiro, Indiana Hemophilia & Thrombosis Center, Indianapolis, IN; SLentz, University of Iowa, Iowa City, IA; JGill, V. Flood, Comprehensive Center for Bleeding Disorders, Milwaukee, WI; CLeissinger, Tulane University Health Sciences Center, New Orleans, LA; MRagni, University of Pittsburgh, Pittsburgh, PA; MTarantino, JRoberts, Bleeding & Clotting Disorders Institute, Peoria, IL; P. James, Queen’s University, Kingston, ON, Canada.

IN ADDITION, NUMEROUS SECONDARY CENTERS CONTRIBUTED TO SUBJECT RECRUITMENT

J. Hord, Akron Children’s Hospital, Akron, OH; J. Strouse, Johns Hopkins Children’s Center, Baltimore, MD; AMa, University of North Carolina Chapel Hill, Chapel Hill, NC; L. Valentino, LBoggio, Rush University Medical Center, Chicago, IL; ASharathkumar, Children’s Memorial Hospital, Chicago, IL; RGruppo, Cincinnati Children’s Hospital, Cincinnati, OH; B. Kerlin, Nationwide Children’s Hospital, Columbus, OH; RKulkarni, Michigan State University, East Lansing, MI; DGreen, Northwestern University, Evanston, IL; K. Hoots, DBrown, University of Texas Health Science Center at Houston, Houston, TX; DMahoney, Baylor College of Medicine, Houston, TX; L. Mathias, ABedros, Loma Linda University Medical Center, Loma Linda, CA; C. Diamond, University of Wisconsin Madison, Madison, WI; A. Neff, Vanderbilt University, Nashville, TN; D. DiMichele and P. Giardina, Weill Cornell Medical College, New York, NY; ACohen, Newark Beth Israel Medical Center, Newark, NJ; M. Paidas, Yale School of Medicine, New Haven, CT; E. Werner, Children’s Hospital of the King’s Daughters, Norfolk, VA; A. Matsunaga, Children’s Hospital & Research Center Oakland, Oakland, CA; F. Shafer, Drexel University College of Medicine, Philadelphia, PA; BKonkle, ACuker, University of Pennsylvania, Philadelphia, PA; P. Kouides, Rochester General Hospital, Rochester, NY; D. Stein, Toledo Children’s Hospital, Toledo, OH.