The data listed in the CHBMP mutation list were collected from the current F9 mutation databases combined with a systematic literature search. The mutations in existing public access F9 mutation databases including the FIX mutation database (http://www.factorix.org/), its previous version, HGMD (http://www.hgmd.cf.ac.uk/), and UniProt (http://www.uniprot.org/) were identified. The cited publications for each mutation in these databases were reviewed for accuracy. A systematic literature search using PubMed (http://www.ncbi.nlm.nih.gov/pubmed) was conducted for publications that reported F9 mutations not listed in the current databases using the following terms: “Hemophilia B and mutation,” “F9 and mutation,” and “factor 9 and mutation.” Finally, mutations that were identified in the CDC laboratory as part of the Hemophilia Inhibitor Research Study (HIRS) (Miller et al. 2012) that were not published were added. Quarterly PubMed searches will be conducted to identify novel mutations for CHBMP updates.

Data were compiled as described in the CDC CHAMP mutation list (Payne et al. 2013). Briefly, only single mutations that potentially cause HB were included in the CHBMP mutation list. To avoid phenotypic ambiguity, mutations in female carriers and patients carrying multiple disease-causing mutations were excluded. Similarly, polymorphisms were excluded because they are not associated with the HB phenotype. Each mutation was assigned a name for cDNA according to the nomenclature recommendations of the HGVS (http://www.hgvs.org/mutnomen/) based on the reference sequence NM_000133.3. The corresponding positions of the mutations in the human genome version hg19 and in the widely used nucleotide sequence described in Yoshitake et al. (1985) were also provided. When available, predicted changes in amino acids caused by the mutations were named using HGVS nomenclature based on protein reference sequence NP_000124.1. The widely used mature protein nomenclature, in which the numbering started at the first amino acid of the processed protein, was listed as well (Yoshitake et al. 1985). Each mutation was also provided with additional information including: the exonic, codon and domain location of the mutation; the mutation type and the underlying mechanism; the disease severity associated with each mutation; the inhibitor history; and the reference that first reported the mutation and the year published.

As described in Payne et al. (2013), the mutation types were classified into nine categories: missense, nonsense, frameshift, synonymous, promoter, 3′UTR, small structural changes (in frame; <50 bp), splice site change, and large structural changes (>50 bp). The molecular defects resulting in these mutation types were listed under Mechanism. These included: substitution, deletion, duplication, insertion, and deletion/insertion, if a number of base pairs were deleted and extra base pairs were inserted in the same place in F9. Exon and codon identify the exons where the mutation is located and codons the mutation affects, respectively. The domains, which were defined based on Chavali et al. (2011), identify the domain locations of the mutation. The subtype of each mutation indicates the mutation locations in one of the following: promoter, signal peptide, propeptide, light chain, heavy chain, and 3′UTR.

According to the recommendation of the Scientific and Standardization Committees of the International Society of Thrombosis and Haemostasis (ISTH) (White et al. 2001), the severity of each mutation was classified based on FIX activity levels as severe (<1%), moderate (1–5%), and mild (>5 to <40%). Reported severity listed the severity reported in the original articles.

The history of inhibitors of each mutation was listed if this was reported from any of the following sources: the original publication, the current HB database, or record in the CDC HIRS database. Otherwise, Not Reported was listed. Each mutation was provided with a reference that first reported the mutation. If a mutation was not published in the literature but was in the FIX mutation database (http://www.factorix.org/) or was identified in HIRS, it was cited as FIX mutation databases or CDC, respectively. The year of the publication was listed in a separate field. Additional information such as potential functional consequences of the mutation, whether or not the mutation occurs at a CpG site, and the dbSNP polymorphism number, if applicable, were listed under “Comments.”

The CHBMP mutation list is developed in the format of a Microsoft Excel file. There are eight tabs in the Excel workbook. Table of Contents summarizes the contents of each tab. The CHBMP mutation list tab contains all the information of F9 mutations described above. The Field Definition tab provides explanations for the fields that are used to describe each mutation. There are six figures in the Figures tab to summarize the data in the CHBMP mutation list. The Tables tab contains two tables that summarize the frequency of mutation types of F9, inhibitor information, and the associated phenotypes. All the references cited in the CHBMP mutation list are listed in the References tab. A search guide is provided in the Instructions tab. A form for mutation submission and correction is provided in the Submission tab which allows users to provide information for updates and corrections.

The frequencies of the nine mutation types are outlined in Table 1. The most common mutation in F9 was missense, which accounted for 58.4%. A smaller percentage (15.4%) of the mutations was frameshifts caused by deletion, duplication, or insertion. Nonsense (8.3%) and splice site (9.4%) mutations were common in the population as well. Other mutations, large structural change (3.0%), small structural change (1.9%), synonymous mutation (0.9%), and mutations in promoter (2.2%), or 3′UTR (0.4%) accounted for a small percentages of listed mutations.

The development of neutralizing antibodies, termed inhibitors, has been reported in 1.4–11.7% of HB patients and in 3.7–24% of severe cases (Radic et al. 2013). It has also been reported that more than 20% of patients with gross deletions develop inhibitors (DiMichele 2007). For most of the mutations in the mutation list, inhibitor status was not reported in the original publications. If inhibitors were reported in any patient carrying a mutation, the mutation was counted as being associated with inhibitors. Only 136 (12.6%) mutations had inhibitor status reported. Twenty four of them, including nine large deletion, two missense, six nonsense, six frameshift, and one splice mutations, were associated with inhibitors.

Figure 2 shows the distribution of frameshift, missense, and nonsense mutations throughout the FIX protein. All three types of mutations are represented in every domain of FIX; however, the majority of the mutations are located in the protease domain (Fig. 2). The map of the mutations on FIX according to mutation type is shown in Figure 3 with each bar representing the number of mutations per 10 codons. Frameshift and nonsense mutations are evenly distributed throughout the FIX protein. This is expected as these types of mutations may cause premature stop of translation, thus leading to defective or truncated proteins regardless of their location. Missense mutations are present in all domains except the middle region of the activation domain. Consistent with previous reports (Giannelli et al. 1998), signal peptide and activation domains are underrepresented among missense mutations due to the lack of importance of most of their amino acids. Because the activation domain is cleaved off during the process of activation, it is less functionally important. However, missense mutations are found in the junction regions between the activation domain and its adjacent domains suggesting that the mutation might interfere with the normal cleavage process and result in the production of dysfunctional FIX proteins.

Large structural changes, including deletions and insertions involving more than 50 bp, often have deleterious effect and are associated with severe phenotype. There are 33 different deletions and insertions in the CHBMP mutation list, 29 of which are large deletions ranging from single-domain deletions to whole-gene deletions. Notably, there are two large deletions spanning two noncontinuous regions of FIX (Fig. 4). One issue is that the exact break points for the majority of large deletion mutations are not reported in the literature. We have distinguished them at the exon level. Therefore, the true number of unique large deletions is likely greater than that in our list.

The severity reported in the original publication is listed in a separate column in the CHBMP mutation list. Some of the mutation types exhibited variability in severity among different patients having the same mutation type. This is due to the fact that severities were classified in some publications using a different standard than that currently adopted by the ISTH. Here, we summarize the distribution of severity for the mutation types based on the ISTH standard (Table 2). The severity was reported for 902 mutations (83.3%) of the total 1083 mutations listed. A total of 194 mutations (21.5%) were reported with inconsistent severities. The majority of the mutations (442, 49%) were associated with the severe phenotype only. This is the case for most of the mutation types except mutations in the promoter and 3′UTR. The moderate phenotype only was reported for 147 mutations (16.3%), and the mild phenotype only for 119 mutations (13.2%). The vast majority of frameshift (78.5%), nonsense (75.3%), and large structural change mutations (90%) were associated with the severe phenotype. This is expected as these mutations result in fundamental changes in the FIX protein. Mutations in the promoter may affect the transcription efficiency and result in a low level of FIX as well as the Leyden phenotype. Only 16.7% of the mutations in the promoter led to severe phenotype. None of the mutations in 3′UTR caused severe HB (Table 2).

The mutations associated with mild, moderate, and severe phenotypes are distributed evenly throughout F9 (Fig. 5). The distribution patterns are similar among the three categories of phenotypic severity. Mutations associated with severe phenotype account for the majority in all domains of FIX. It was noted that neither mild nor moderate phenotype was reported within a small part of the activation domain. This is likely due to the fact that the domain is cleaved in the course of activation. Mutations identified in this region were either frameshift or nonsense, which mainly causes severe phenotype.