The Canadian “National Program for hemophilia mutation testing” database: A ten-year review
This article is corrected by:
- Errata: The Canadian “National Program for hemophilia mutation testing” database: A ten-year review Volume 89, Issue 6, 669, Article first published online: 20 May 2014
A reference genotyping laboratory was established in 2000 at Queen's University, Kingston, to provide genetic testing for Hemophilia A (HA) and B (HB) and create a Canadian mutation database. Canadian hemophilia treatment centers and genetics clinics provided DNA and clinical information from November 2000 to March 2011. The factor VIII (F8) gene was analyzed in 1,177 patients (47% of HA population) and 787 female family members and the factor IX (F9) gene in 267 patients (47% of HB population) and 123 female family members, using Southern Blot, PCR, conformation sensitive gel electrophoresis, and/or direct sequencing. The mutation detection rates for HA and HB were 91% and 94%, respectively. 380 different F8 mutations were identified: inversions of intron 22 and intron 1, 229 missense, 45 nonsense, eight deletions, 70 frameshifts, 25 splice site, and one compound mutation with a splice site and intron 1 inversion. Of these mutations, 228 were novel to the Hemophilia A Database (HADB, http://hadb.org.uk/). A total 125 different F9 mutations were identified: 80 missense, 12 frameshift, 12 splice site, nine nonsense and seven promoter mutations, three large deletions, and two compound mutations with both missense and nonsense changes. Of these mutations, 36 were novel to the International Haemophilia B Mutation database (http://www.kcl.ac.uk/ip/petergreen/haemBdatabase.html). The Canadian F8 and F9 mutation database reflects the allelic heterogeneity of HA and HB, and is similar to previously described populations. This report represents the largest and longest duration experience of a national hemophilia genotyping program documented, to date. Am. J. Hematol. 88:1030–1034, 2013. © 2013 Wiley Periodicals, Inc.
Hemophilia A (HA) and B (HB) are X-linked recessive bleeding disorders caused by mutations in the F8 gene or F9 gene, which result in deficient or defective coagulation factors. The two disorders represent the most common of the severe inherited coagulation disorders with a combined prevalence of approximately 1 in 10,000 [1, 2]. F8 is a large gene (186 kb) that is found in the most distal band (Xq28) of the long arm of the X chromosome . F9 is also located on the long arm of the X chromosome at Xq27 and is 34 kb in length . Both HA and HB are marked by significant allelic heterogeneity. More than 1,200 unique mutations have been identified for HA and have been entered into the Hemophilia A Database (HADB, found at http://hadb.org.uk/). HB mutations have been compiled on the International HB Mutation Database, found at http://www.kcl.ac.uk/ip/petergreen/haemBdatabase.html.
Identification of the underlying F8 or F9 mutation is important for optimal family counseling, prompt carrier and prenatal diagnosis, and prediction of risk for inhibitor formation or anaphylactic reaction following exposure to infused replacement therapy . The high degree of mutational heterogeneity complicates mutation testing. To optimize access to genetic services in Canada, a national genotyping laboratory and mutation database was established in November 2000 at Queen's University, Kingston. Here, we review 10 years of the reference laboratory's activity.
Patients, clinical information
The establishment of the “National Program for Hemophilia Mutation Testing” was reviewed and approved by the research ethics board at Queen's University, Kingston, Canada, and special approval for the publication of the anonymous legacy data was obtained. Since November 2000 to March 2011, samples and a requisition, outlining basic clinical information were sent by any one of the 26 Canadian Hemophilia treatment centers (HTCs) or clinical genetics clinics throughout Canada. Rarely, in remote areas of Canada, general practitioners also sent in samples.
The requisitions used are available at http://clinlabs.path.queensu.ca/queens/labs/lillicrap/gl.htm. The clinical information requested includes sex, date of birth, HA or HB, baseline coagulation factor level, presence of an inhibitor, inhibitor titre, family history, and reason for testing. Many of the requisitions were received incomplete. Between June and December 2011, HTCs were approached to provide the missing information, particularly factor level, and history of an inhibitor.
In brief, the approach to mutation testing involved ruling out intron 22 inversion and intron 1 inversion with southern blotting and PCR in cases of severe HA. All samples received before 2009 were then screened with conformation sensitive gel electrophoresis (CSGE) and the abnormal CSGE PCR product was sequenced. If the CSGE failed to identify an abnormal PCR product, or if the sample was received after January 2009, the entire coding sequence was amplified. Finally, mutation-negative cases of low FVIII were then earmarked for type 2N VWD mutation screening.
Genomic DNA was isolated from anticoagulated peripheral blood samples by the standard salting out procedure .
Southern blotting and PCR
For the detection of the intron 22 inversion, Southern Blots of BclI digested DNA were carried out as previously described using the 0.9kb EcoRI/SstI fragment from plasmid p482.6 (catalogue no. 57203, ATCC, VA) as a hybridization probe . Intron 1 inversions were detected by PCR, as previously reported .
Screening with CSGE
Screening with CSGE was used up to January 2009, after which all patient samples were subjected to direct sequencing. For details regarding the CSGE method please see Refs. 9,10.
In cases screened with CSGE, the abnormal CSGE PCR product was sequenced. If the CSGE failed to identify an abnormal PCR product, or if the sample was received after January 2009, the entire coding sequence was amplified: including all 26 exons of F8 gene and 8 exons of F9 gene, intron/exon boundaries, the 5′ and 3′ untranslated regions and the proximal promoter regions. Using a DNA thermal cycler (Perkin Elmer Life Sciences, Shelton, CT), a T3 Thermocycler (Biometra, Montreal, QC, Canada), or an ABI GeneAmp (Applied Biosystems, Foster City, CA), DNA was amplified for 30–35 cycles of 45–60 seconds at 94°C, 45–60 seconds at 53–65°C, and 45–60 seconds at 72°C. The amplified products were purified using the QIAGEN QIAquick PCR Purification Kit and sequenced on an ABI model 373 automated sequencer. Once a plausible mutation was identified, no further sequence analysis was performed. All nucleotide changes were confirmed by repeating the PCR and the opposite DNA strand was sequenced to confirm the sequence variation.
Splice site prediction
Mutations at or near splice junction consensus sequences were analyzed by the Berkeley Drosophila Genome Project (intron) and ESE finder version 3.0 (exon) to predict changes in RNA splicing.
Sequencing for type 2N VWD
The VWF FVIII-binding assay is not available in Canada. Cases with low FVIII levels but with no F8 mutation were investigated for type 2N VWD by sequencing of exons 17–21, and 24–27 of the VWF gene .
The cDNA numbering system and protein reference sequence for both F8 and F9 mutations is in accordance with the Human Genome Variation Society recommendations . Aliases using the nomenclature common to HADB or the International HB mutation database are also included in the tables in parentheses.
Population of HA patients
We genotyped 1,177 males and 787 females. Indications for testing are summarized in Table 1. 5% (n = 105) of all F8 samples were urgently requested during pregnancy or for prenatal diagnosis. Information regarding gestational age and outcomes is not available. Our mutation detection rate is 91% (1,074 of 1,177 male patients). The frequencies of different F8 mutations based on clinical phenotype are summarized in Table 2. FVIII inhibitor status was reported in 1,068 of 1,177 patients, with 82 (8%) having a history of a FVIII inhibitor: 58 (of a total 395 with known inhibitor status, 15%) cases were in severe HA and eight (of a total 163, 5%) in moderate HA, 14 (of a total 465, 3%) in mild HA and one (of a total 35, 3%) in a patient with unknown levels. FVIII inhibitor was reported in one female with heterozygosity for the missense mutation (1,621 A>T, Thr522Ser) and unreported baseline factor level. (Table 3) The mutations associated with the inhibitors are summarized in Supporting Information Table SI. Missense mutations are generally low risk for inhibitor development , but five mutations (A612C, Y2124C, R2169H, Y2248C, P2319L) have been associated with a 23–50% risk of inhibitor development . Our experience with these missense mutations is summarized in Table 4. Three hundrend and eighty unique F8 mutations were identified: in addition to intron 22 and intron 1 inversions, 229 missense mutations, 45 nonsense, eight deletions, 70 frameshifts, 25 splice site, and one compound mutation with both a splice site change as well as an intron 1 inversion. Excluding exon 14, point mutations were distributed throughout the gene with 2–24 unique mutations identified per exon and loosely correlating to exon size. Exon 14, encoding the B domain, is the largest exon, accounting for 43% of the coding region (3,106 of 7,227 bp). Twenty-one % (65 of 317) of all point mutations occurred in exon 14 but only 9% (21 of 229) of the missense mutations. As the B domain lacks procoagulant activity, this previously reported finding likely reflects that missense mutations within this region do not cause HA [14, 15]. However, exon 14 is a hotspot for small deletions and/or insertions . Forty-seven % (33/70) of frameshift mutations occurred in exon 14. Single nucleotide insertions or deletions into one of the two runs of A nucleotides at codons 1,191–1,194 or 1,439–1,441 have been reported to be frequently occurring mutations . In our population, these mutations accounted for 41% (55 of 133) of the cases with frameshift mutations. The most frequently identified mutation is the intron 22 inversion in 129 severely affected males. The second most frequently identified mutation is the missense mutation, V2035A (V2016A in the “old” HADB nomenclature) in exon 19, found in 42 individuals from an isolated population in Newfoundland and is likely because of a founder effect . Finally, 228 F8 mutations that were novel to the HADB are summarized in Supporting Information Table SII.
Table 1. Indication for Genotyping F8 and F9
|None identified||343 (29)||129 (16)||91 (34)||15 (12)|
|Documented Family History||289 (25)||36 (5)||68(25)||9 (7)|
|New/Sporadic Case||439 (37)||16 (2)||98 (36)||3 (2)|
|Unusual Case for research||30 (3)||15 (2)||10 (4)||1 (1)|
|Carrier testing||69 (6)||493 (63)||N/A||75 (61)|
|Prenatal testing or testing during pregnancya||7 (1)||98 (12)a||0||20 (16)a|
Table 2. Distribution of F8 and F9 Mutation Types According to Clinical Phenotype
|No mutation identified||103 (9)||20 (5)||9 (5)||51 (10)||23 (31)|
|Intron 22 Inversion||129 (11)||129 (31)a||0 (0)||0 (0)||0 (0)|
|Intron 1 Inversion||7 (1)||7 (2)||0 (0)||0 (0)||0 (0)|
|Missense||670 (57)||65 (16)||113 (67)||448 (86)||44 (59)|
|Nonsense||70 (6)||54 (13)||13 (8)||0 (0)||3 (4)|
|Large Deletion||20 (2)||18 (4)||1 (1)||1 (<1)||0 (0)|
|Frameshiftb||129 (11)||100 (24)||22 (13)||2 (<1)||5 (7)|
|Splice Site||37 (3)||17 (4)||8 (5)||12 (2)||0 (0)|
|Other (includes ≥2 mutations, VWD or Hereditary parahemophilia)||12 (1)||3 (1)||2 (1)||7 (1)||0 (0)|
|No mutation identified||18 (7)||1 (1)||0 (0)||9 (16)||7 (22)|
|Missense||174 (65)||59 (57)||62 (83)||36 (64)||18 (56)|
|Nonsense||24 (9)||18 (17)||3 (4)||0 (0)||3 (9)|
|Large Deletion||4 (1)||4 (4)||0 (0)||0 (0)||0 (0)|
|Frameshift||15 (6)||12 (12)||2 (3)||1 (2)||0 (0)|
|Splice Site||17 (6)||7 (7)||4 (5)||4 (7)||2 (6)|
|≥2 mutations||2 (1)||2 (2)||0 (0)||0 (0)||0 (0)|
|Hemophilia B Leiden||13 (5)||1 (1)||4 (5)||6 (11)||2 (6)|
Table 3. Rates of inhibitors as per Hemophilia Phenotype
|All||82/1068, 8%||4/246, 2%|
|Severe||58/395, 15%||3/103, 3%|
|Moderate||8/163, 5%||1/76, 1%|
Table 4. Missense Mutations Associated with a High Risk of Inhibitor Development: The Canadian experience
|A612C (A593C)||27 (29)||2 (7)||1 Severe, 1 Moderate, 22 Mild, 5 Unknown|
|Y2124C (Y2105C)||3 (3)||3 (100)||2 Mild, 1 Moderate|
|R2169H (R2150H)||27 (27)||4 (15)||2 Severe, 14 Moderate, 10 Mild|
|Y2248C (Y2229C)||4 (4)||1 (25)||2 Moderate, 2 Mild|
|P2319L (P2300L)||21 (24)||0 (0)||24 Mild|
For the females, F8 mutations were identified in 485 of 787 samples received. F8 mutations were identified in 85% (n = 123) of symptomatic females: mutation detection rates based on phenotype were 100% for females with FVIII levels <1% (n = 3), 80% with levels of 1–5% (n = 5) and 85% with levels of >5–50% (n = 115). All three females with FVIII levels <1% were heterozygous for intron 22 inversion, but a second mutation which would explain the severe phenotype was not identified despite full F8 sequencing.
Although all cases with low FVIII levels and no F8 mutation identified will eventually be subjected to type 2N VWD genotyping, only a subset has been processed thus far: 38 samples received from males and 40 samples received from females. Type 2N VWD was diagnosed in 24% (n = 9) of the males and 20% (n = 8) of the females. Six patients were homozygous for the mutations R854Q (n = 5) and C858S (n = 1), two patients were found to have a compound heterozygous mutation (R816Q/R854Q), and 9 patients had heterozygous mutations identified (8 with R854Q and 1 with R816Q).
Population of HB patients
Two hundred and sixty seven males and 123 females were genotyped for F9 mutations. Indications for testing are summarized in Table 1. 7% of all F9 samples submitted were required urgently for prenatal diagnosis. F9 mutations were identified in 250 of 267 patients, for a detection rate of 94%. The frequency of F9 mutations based on clinical phenotype is summarized in Table 2. FIX inhibitor status was reported in 246 of 267 patients, with four (2%) patients having a history of a FIX inhibitor: three cases were in severe HB and one in moderate HB. The mutations associated with these inhibitors were, a deletion of exons A–D, deletion of exons E–H, a nonsense mutation (R294Stop), and a missense mutation (A266V). One hundred and twenty five unique F9 mutations were identified: 80 missense mutations, 12 frameshift, 12 splice site, nine nonsense, seven promoter mutations causing HB Leiden, three large deletions, and two compound mutations with both missense and nonsense changes. Point mutations were distributed throughout the gene with 45% of mutations occurring in the catalytic region of the gene (exons G–H), encoded by 235 of the protein's 415 amino acids. Thirty-six of the F9 mutations (Supporting Information Table S3) were novel to the International HB Mutation database (http://www.kcl.ac.uk/ip/petergreen/haemBdatabase.html) at the time of discovery.
For the females, F9 mutations were identified in 79 of 123 samples received. F9 mutations were identified in 100% (n = 21) of symptomatic females: one female with a FIX level between 1–5% was heterozygous for a missense mutation and 20 females with FIX levels >5–50% had 15 missense, one nonsense, and four splice site mutations.
The activity of the mutation testing program
The rate of requests for genotyping has remained stable for F9 with 31 to 46 samples/year. F8 genotyping requests previously ranged from 132 to 204 samples/year, but the years 2009 and 2010 have seen an increase to 243 and 268 samples/year, respectively. All individuals in Canada with hereditary bleeding disorders are registered in the Canadian Hemophilia Registry (CHR), which is updated annually (available at http://fhs.mcmaster.ca/chr/index.html, last updated May 7, 2011). We have compared the number of individuals genotyped to the number of registered patients (see Table 5).
Table 5. The Number of Hemophilia Patients in Canada as per the Canadian Hemophilia Registry (CHR) Compared to the Numbers that have been Genotyped at the Reference Laboratory
We have genotyped 1,964 individuals for F8 and 390 individuals for the F9 mutations. The experience of this testing facility is representative of a clinically driven hemophilia genotyping program. Comparison of the population of HA and HB patients genotyped with the CHR (Table 5), reveals some bias towards severe/moderate HA patients and severe HB patients for genotyping. Severely affected patients are seen more frequently in clinic, and have more opportunity to be tested. Also, more severely affected patients are at generally higher risk of inhibitor development ; knowledge of the underlying mutation type may be helpful to further stratify inhibitor development risk. The mutation detection rates of 91% and 94% for HA and HB, respectively, are comparable to several previously published reports of mutation databases [5, 14, 19-21], but lower than predicted in view of current gene sequencing technology [22-24]. Several factors may be impacting our mutation detection rates. First, until early 2009, the reference laboratory used CSGE for mutation screening with an expected mutation detection efficiency of 85–90% . The mutation negative cases from the era of CSGE screening are being reassessed using F8 and F9 direct sequencing with expected detection rates of >95%. Second, our algorithm will miss large duplications that may account for 11–17% of the mutation negative cases [25, 26]. Methods, such multiplex ligation-dependent probe amplification (MLPA), which will identify large duplications within F8 and F9 genes in hemophilia patients, and large duplications or deletions in hemophilia carriers have not yet been routinely adopted by our program [25, 27]. Third, several of the samples were of inadequate quality or quantity: additional whole blood or DNA has been requested. Finally, as a result of the broad referral base of the genotyping laboratory, some patients will not have the specified diagnosis. For example, 32 patient samples sent for F9 genotyping and 75 samples sent for F8 genotyping had unknown coagulant levels. These patients may have been misdiagnosed as hemophilia: at least six mutation negative patients were subsequently identified as having VWD or a different coagulation deficiency to that which had originally been specified. F8 mutation negative cases with abnormal FVIII levels are triaged to be genotyped for VWD type 2N: 22% (n = 17) of 78 patients genotyped for type 2N VWD were found to have causative mutations, with eight being homozygous or compound heterozygous, and nine being heterozygous. The heterozygous cases likely represent compound heterozygosity with a coexistent VWF null mutation although this was not confirmed with genotyping. On the other hand, type 1 and 3 VWD could be missed in our algorithm. Mutation negative cases of mild or moderate hemophilia are not being routinely subjected to genotyping for combined FV/FVIII deficiency. The patient groups with unknown coagulant levels account for 22% (23 of 103) and 33% (7 of 21) of mutation negative cases for HA and HB, respectively.
As novel HA and HB mutations were identified in this program, they were submitted to the HADB and the International HB Mutation database up to mid 2005, the time of the last major update for these databases. Nonetheless, these mutations have been included in Supporting Information Tables SII and SIII to provide a more comprehensive representation of the novel mutations found in the Canadian hemophilia population at the time of discovery. Thirty-six and 228 of mutations were novel at the time of discovery to the International HB Mutation Database and to HADB, respectively. Missense mutations accounted for 21/36 of novel HB and 135/231 of novel HA mutations. A potential deleterious missense mutation was distinguished from a polymorphism by interrogation of the SNP database found on HADB (http://hadb.org.uk/) or on the International HB Mutation Database (http://www.kcl.ac.uk/ip/petergreen/haemBdatabase.html). Confirmation of the mutation in affected family members when possible supported its being a causative change. No additional in silico analysis of the functional consequences of these changes was undertaken, and expression studies are outside the scope of this study. Three F8 missense mutations that had not been previously identified were seen with increased frequency. V2276A in exon 25 was identified in 17 moderate or mild HA males and 8 heterozygous females. Sixteen of these individuals belonged to two families. The second missense mutation, R2135G in exon 22 was seen in 10 moderate and mild HA males, one homozygous mild HA female and three carriers. Ten of these individuals belonged to 4 families. Finally, V115A in exon 3 was identified in 8 mild HA males and seven female carriers, all of whom belonged to five families. These samples were predominantly sent from HTCs in Eastern Canada (Halifax, Moncton and Newfoundland) and almost certainly represent a founder effect.
The mutation spectrum for HB reported here is consistent with previously published population studies [19, 21, 28]. However, the frequency of certain HA mutation will differ from previous reports [14, 15, 20]. Two large Canadian HTCs perform intron 22 inversion testing prior to sending samples for further analysis. Therefore, our reported intron 22 inversion mutation rates underestimate the contribution of this mutation to the Canadian severe HA population. In contrast, certain mutations, for example the previously mentioned V2035A in exon 19 found in 42 individuals, are over-represented because of a founder effect. One severe HA patient and two severe HB patients were found to have more than one mutation: the intron 1 inversion and a splice site mutation for the HA patient, and a missense and nonsense mutation for both of the HB patients. Further information regarding additional polymorphisms is not available. These three cases represent exceptions: generally, cases in which there may be more than one sequence variant will be missed as once a plausible mutation was identified, no further genetic analysis was performed.
Clinical information regarding baseline factor levels and history of inhibitors must be interpreted with caution. Requisitions were completed by a variety of health care professionals ranging from hematologists, to medical office assistants. Furthermore, many of the requisitions received were incomplete. Between June and December 2011, HTCs were contacted regarding clinical information that was not completed on the original requisitions, particularly baseline factor level, history of an inhibitor (yes/no) and peak inhibitor titre. Our inhibitor numbers reflect a cumulative incidence estimate of this treatment complication. No information regarding frequency of inhibitor testing, type of inhibitor assay used or the cut-off values for positive tests is available. Comparison of our inhibitor rates to other populations show that our inhibitor rates are generally lower, but it must be stressed that because of the nature and conduct of this analysis it is highly likely that our numbers underestimate the true inhibitor prevalence in the Canadian hemophilia population [5, 14, 20, 29].
This manuscript represents the largest and longest duration experience yet reported for a national hemophilia genotyping program. The national database enables the analysis of the mutation spectrum and lays the ground work for genotype–phenotype correlations, as well as the determination of recurrent mutations that have resulted from a founder effect. As the underlying hemophilia mutation represents the most significant risk factor for inhibitor development [5, 30, 31], knowledge of a high or low risk mutation will impact management. Finally, while genotyping of approximately 50% of the HA and HB populations has now been accomplished in Canada, the demand for this service has remained constant, and even increased for HA testing. This trend emphasizes the growing realization that knowledge of the hemophilia mutation can aid in the application of a personalized medicine approach to the management of these patients.
The Canadian National Program for Hemophilia Mutation Testing has been supported with funds from Health Canada, Baxter Bioscience and most recently, by the Association of Hemophilia Clinic Directors of Canada. David Lillicrap is the recipient of a Canada Research Chair in Molecular Hemostasis. Natalia Rydz is the recipient of a Bayer Hemophilia Clinical Training Award.
Conflict of interest
The authors have no interests which might be perceived as posing a conflict or bias. JL and ST performed the genotyping. NR wrote the paper. PJ and DL edited the paper and are the directors of genotyping laboratory.