Molecular pathogenesis and therapeutic advances in haemophilia, an update of the current evidence

Abstract

Haemophilia A and B are X-linked inherited bleeding disorders caused by quantitative or qualitative deficiencies of coagulation factors VIII and IX, respectively. Understanding the underlying genetic pathogenesis and its variants is of paramount importance for current treatment targets, such as gene therapy. The current comprehensive review synthesizes the current evidence of molecular pathogenesis of haemophilia A and B, while it presents the therapeutic strategies, their limitations and future directions. Advances in molecular genetics have defined a highly heterogeneous mutational spectrum. Recurrent F8 inversions (intron 22 and 1) dominate severe phenotypes, while null mutations (nonsense, frameshift) are strongly associated with increased inhibitor risk. Conversely, missense variants predominantly underlie mild-to-moderate forms, and the identification of gain-of-function variants has directly informed next-generation therapeutic design. Clinically, strategies have transitioned from episodic replacement to individualized prophylaxis with ultra-long-acting concentrates and non-factor mimetics (emicizumab), which effectively prevent joint disease even in the presence of inhibitors. Furthermore, adeno-associated viral-mediated gene therapy offers a path toward a functional cure through sustained endogenous factor expression, although the long-term durability of FVIII expression and hepatic safety remain the primary current challenges. This review bridges the gap between the complex molecular landscape of haemophilia and the latest therapeutic breakthroughs. By linking specific genotypes to clinical outcomes and inhibitor risk, it provides a framework for precision management using factor, non-factor, and gene-based strategies in the current clinical landscape.

Key Words: Haemophilia A; Haemophilia B; F8 gene; F9 gene; Molecular pathogenesis; Gene therapy; Adeno-associated virus; Non-factor replacement therapy; Emicizumab; Prophylaxis

Core Tip: The current comprehensive review synthesizes the current evidence of molecular pathogenesis of haemophilia A and B, while it presents the therapeutic strategies, their limitations and future directions. Advances in molecular genetics have defined a highly heterogeneous mutational spectrum in F8 and F9, including recurrent inversions, point mutations, small insertions or deletions, and large rearrangements. These variants underlie genotype-phenotype correlations. Therapeutic strategies have evolved from on-demand plasma-derived factor replacement to individualized prophylaxis with standard and extended half-life concentrates. Parallel progress in liver-directed adeno-associated viral gene therapy has enabled sustained endogenous expression of FVIII and FIX following a single infusion, with marked reductions in annualized bleeding rates and factor use.

INTRODUCTION

Haemophilia comprises a group of inherited bleeding disorders caused by quantitative or qualitative deficiencies of coagulation factor VIII (haemophilia A) or factor IX (haemophilia B). Both disorders follow an X-linked recessive inheritance pattern and affect mostly males, while females are typically asymptomatic carriers. However, some carriers may exhibit clinically significant bleeding due to low factor levels[1]. Haemophilia A is the more common type, with an estimated prevalence of approximately 1 in 5000 male births, whereas haemophilia B occurs in about 1 in 30000 male births[2]. Landmark meta-analysis suggest that the true prevalence of the disorder may be substantially higher than historically assumed, with pooled prevalence estimates of 17.1 per 100000 males for haemophilia A and 3.8 per 100000 males for haemophilia B[3]. However, in 2026, longitudinal data indicate a significant upward trend in reported cases globally[4]. According to the latest global projections, reported prevalence has increased by nearly 36% over the last decade, driven by enhanced diagnostic capabilities in emerging regions and improved survival rates in high-income countries[4,5].

The clinical phenotype of haemophilia is mainly defined by the residual plasma activity of FVIII or FIX. International consensus definitions, endorsed by the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis, classify haemophilia into three severity categories based on the factor activity levels: Severe: < 1% (< 0.01 IU/mL), moderate: 1%-5% (0.01-0.05 IU/mL) and mild: 5%-40% (0.05-0.40 IU/mL)[6]. This classification is able to predict bleeding tendency and is used to guide therapeutic decisions. Although these categories typically reflect the bleeding risk with precision, there is substantial clinical heterogeneity, particularly among individuals with severe haemophilia A. This heterogeneity highlights the influence of underlying molecular defects and additional modifying factors[6].

Pathophysiologically, these deficiencies disrupt the intrinsic pathway of the coagulation cascade, severely impairing the amplification of thrombin generation on the surface of activated platelets. This failure of durable clot formation leads to the clinical manifestations that correspond to spontaneous or prolonged trauma-induced haemorrhages, particularly into large synovial joints and muscles[1]. Thus, haemophilia is characterized by a lifelong predisposition to spontaneous and trauma-related bleeding, with muscle and joint haemorrhages constituting the hallmark of the disease severity. Recurrent haemarthroses, particularly into the ankles, knees, and elbows, initiate a progressive cycle of synovial inflammation, cartilage destruction, and joint deformity. This will ultimately result in haemophilic arthropathy, one of the most disabling long-term consequences of the disorder[7]. Beyond the musculoskeletal complications, haemophilia carries a substantial risk of life-threatening haemorrhage. Intracranial haemorrhage (ICH) represents the most catastrophic bleeding event and remains a major cause of morbidity and mortality across all age groups. A recent systematic review involving over 54000 patients reported pooled ICH incidence rates of 2.3 per 1000 person-years in all ages and 7.4 per 1000 person-years in patients under 25 years, with neonates exhibiting a cumulative incidence of roughly 2% per 100 live births[8].

The introduction of prophylactic factor replacement has transformed the clinical outcomes, particularly when initiated early in life. Seminal Swedish experience spanning over 25 years demonstrated that maintaining the minimum factor level above 1% can effectively prevent joint disease and enable normal activity levels[9]. Subsequent randomized trials confirmed that regular FVIII infusions dramatically reduced joint bleeds and prevented structural joint damage compared with episodic treatment[7]. Prophylaxis is now the standard of care in high-resource settings and has been shown to improve quality of life, preserve musculoskeletal integrity, and reduce long-term disability. A central therapeutic challenge remains the development of inhibitors, alloantibodies directed against the infused FVIII or FIX, which render the replacement therapy ineffective. Inhibitors affect approximately 5%-7% of the overall haemophilia A population but occur far more frequently (up to 30%) in the severe form[10].

The rapid clinical integration of gene therapies and non-factor mimetics over the last years has created a critical need for an updated synthesis of the available evidence. Therefore, the aim of this review is to synthesize current knowledge on the molecular underpinnings of haemophilia, delineate genotype-phenotype relationships, critically evaluate the advances in factor and non-factor replacement therapies and provide an updated overview of gene therapy and genome-editing strategies.

MOLECULAR PATHOGENESIS AND GENOTYPE-PHENOTYPE CORRELATION

Haemophilia arises from pathogenic variants in two large, structurally and functionally complex genes: F8, that encodes the coagulation FVIII, and F9, that encodes the coagulation FIX. Both genes are located on the X chromosome and demonstrate the typical pattern of X-linked recessive inheritance. At the molecular level, the pathology of haemophilia is defined by a failure in the amplification phase of the coagulation cascade. FVIIIa serves as a critical non-enzymatic cofactor for the serine protease FIXa. Together, they assemble on the negatively charged phospholipid surface of activated platelets to form the ‘intrinsic tenase’ complex[11]. This complex is responsible for the proteolytic activation of FX to FXa, a step that is roughly 100000 times more efficient than the action of FIXa alone. In the absence of functional FVIII or FIX, thrombin generation is severely attenuated, failing to reach the ‘thrombin burst’ threshold required to convert fibrinogen into a stable, cross-linked fibrin mesh[11].

Their molecular architecture, transcriptional features and mutational landscapes have been extensively studied and described over the past decades, setting the foundation for an up-to-date genetic diagnosis and genotype-phenotype interpretation.

The F8 and F9 genes

The F8 gene is positioned at Xq28 and spans approximately 186 kb, comprising 26 exons. This renders it as one of the largest and most structurally intricate genes in the human genome[11]. FVIII is synthesized as a large multidomain precursor with the organization A1-a1-A2-a2-B-a3-A3-C1-C2 and is encoded by a single long exon (exon 14) that represents about 40% of the coding sequence. Critical structural features include the presence of two intronic gene copies (F8A and F8B) within intron 22, whose homology with upstream extragenic copies predisposes to the well-known intron 22 inversion (Inv22), responsible for 40%-50% of severe haemophilia A cases[11]. Advances in molecular genetics initially included linkage analysis and later direct sequencing. These techniques have enabled a detailed characterization of F8 mutations. The disproportionate frequency of point mutations at CpG dinucleotides is driven by the spontaneous hydrolytic deamination of 5-methylcytosine to thymine (C to T transitions)[11]. Within the F8 gene, these hotspots are particularly consequential when they occur in arginine codons (CGA). For example, a mutation at CGA (arginine) can result in a TGA stop codon, leading to severe haemophilia A due to premature protein truncation[11]. These transitions account for nearly 40% of all single-nucleotide variants in haemophilia A, despite CpG sites representing a negligible fraction of the total genomic sequence[11]. Early work demonstrated the utility of polymorphic markers and RFLPs in order to track the disease allele[12], while subsequent adoption of PCR-based sequencing, long-range PCR for inversion detection, and high-throughput mutational analysis has greatly enhanced the diagnostic sensitivity. The F8 locus is highly susceptible to point mutations at CpG dinucleotides, which account for nearly 40% of small variants despite representing only approximately 2% of the sequence[11]. Modern variant curation efforts, notably the EAHAD FVIII Variant Database, have compiled thousands of variants with detailed annotations, including pathogenicity predictions, structural mapping, and phenotype associations[13]. These curated resources have become essential for interpreting the significance of each variant, assessing the risk of inhibitor development, and refining genotype-phenotype correlations in haemophilia A (Table 1).

Table 1 Genomic and structural characteristics of the F8 and F9 genes of their variant databases.

Feature	F8 gene	F9 gene
Chromosomal location	Xq28	Xq27
Gene size	Approximately 186 kb[11]	Approximately 34 kb[11]
Number of exons	26	8
Protein encoded	Coagulation factor VIII	Coagulation factor IX
Protein structure/domains	A1-a1-A2-a2-B-a3-A3-C1-C2; encoded partly by large exon 14[11]	Signal peptide, propeptide, Gla domain, EGF1, EGF2, activation peptide, serine protease domain[11]
Key genomic features	Intron 22 region containing intronic gene copies (F8A and F8B), predisposing to Inv22 (40%-50% of severe HA)[11]; intron 1 low-copy repeats predisposing to Inv1 (2%-5% of severe HA); CpG hotspots accounting for approximately 40% of small variants despite comprising approximately 2% of sequence[11]	Compact gene with modular exon-domain correspondence; promoter contains androgen-responsive elements responsible for haemophilia B Leyden phenotype[11]
Predominant mutation classes	Inv22 (40%-50% of severe); Inv1 (2%-5% of severe); nonsense, frameshift, splice-site defects dominate in severe HA; missense variants common in mild/moderate HA	Missense (approximately 68%) dominate; nonsense/frameshift variants associated with severe phenotype; splice-site variants relatively frequent; large deletions rare (1%-3%) but high inhibitor risk
Mutation hotspots	Inv22 and Inv1 mediated by homologous recombination	Noncomparable to F8 inversions; mutations distributed across gene
Typical diagnostic methods	Long-range PCR for inversion detection; Sanger/NGS sequencing for point mutations, indels; MLPA for deletions/duplications	Direct sequencing (gene small enough for full coverage); MLPA when deletions suspected

HA: Haemophilia A; Inv22: Intron 22 inversion; Inv1: Intron 1 inversion; NGS: Next gene sequence; MLPA: Multiplex ligation-dependent probe amplification.

The F9 gene, located at Xq27, is considerably smaller than F8, spanning approximately 34 kb with 8 exons that encode a vitamin-K-dependent serine protease structurally similar to factor VII, factor X, and protein C[11]. FIX is synthesized as a precursor protein that contains a signal peptide, a propeptide, one γ-carboxyglutamic acid (Gla) domain, two epidermal growth factor-like domains, and a catalytic serine protease domain. Each is encoded by discrete exons, thus reflecting the modular structure of other vitamin-K-dependent coagulation factors. Molecular diagnostic approaches for haemophilia B have long relied on direct sequencing of the coding region and splice junctions, given the compact size of the gene[12]. The spectrum of F9 mutations differs from that of F8: Missense mutations predominate (approximately 68%), whereas large deletions are substantially less common (approximately 3%), mirroring the lower rate of inhibitor development in haemophilia B[11]. As far as FVIII is concerned, a large proportion of pathogenic point mutations occur at CpG sites. Variants within the promoter region underpin unique phenotypes such as hemophilia B Leyden, which is characterized by childhood deficiency and spontaneous improvement of FIX levels after puberty due to androgen-responsive elements that influence transcription[11]. The EAHAD FIX Variant Database provides a consolidated resource for curated variant information, including the severity grading, the structural annotation, and functional predictions. This data greatly aids clinical interpretation and pathogenicity assessment of each phenotype[13] (Table 1).

Mutational spectrum in haemophilia A

Haemophilia A is characterized by a remarkably heterogeneous mutational landscape within the F8 gene, mainly due to its large genomic size, complex intron-exon structure, and susceptibility to diverse DNA rearrangements. More than 3700 pathogenic variants have been catalogued globally. These include structural rearrangements, point mutations, small insertions or deletions and splice-site defects[14].

The Inv22 is the most frequent genetic lesion in severe haemophilia A and results from intrachromosomal homologous recombination between the int22h-1 region within F8 intron 22 and extragenic int22h-2 or int22h-3 repeats. This rearrangement inverts a 500-600 kb segment of DNA and eventually disrupts the normal transcription of F8. Inv22 accounts for 40%-50% of severe cases in most populations[15-18]. Regional variations are evident: Prevalence may be lower in Albania (10.5% of severe patients)[19] and around 30% in some Indian and Asian cohorts[16,20]. Recent data from northeast India similarly show a 30% prevalence, with a strong enrichment in severe disease[20]. The intron 1 inversion (Inv1) is caused by recombination within intron 1 and represents the second most common recurrent mutation. It is found in approximately 2%-5% of the severe cases[15-17]. These recurrent inversions together represent the only true F8 ‘hotspot’ mutations. Their high prevalence reflects the underlying genomic architecture, especially the presence of low-copy repeats and recombination-prone regions[21].

Beyond recurrent inversions, haemophilia A can be defined by extensive allelic heterogeneity. Multiple studies across diverse populations highlight the predominance of missense mutations, particularly in moderate and mild disease. In severe haemophilia A without inversions, truncating variants including nonsense, frameshift, and canonical splice-site mutations are frequently observed. In Taiwanese patients, 17 different mutations were identified among inversion-negative individuals, including 11 novel variants, mostly missense changes that affected evolutionarily conserved residues[16]. In an Indian cohort, sequencing revealed 17 missense, 5 deletions, 3 insertions, and 2 nonsense mutations, with 11 novel pathogenic variants[17]. A Brazilian study of mild and moderate forms identified 33 different variants, including 27 missense, 1 deletion, 2 duplications, and 3 splice-site mutations, 9 of which were previously unreported[22]. Whole-exome sequencing in a Colombian cohort for the severe form revealed that 70% of variants were premature truncation mutations (nonsense or frameshift), whereas only 18% were missense[14]. These findings consistently demonstrate that null mutations (nonsense, frameshift and essential splice-site) dominate in severe haemophilia A, whereas missense variants mostly concern moderate and mild phenotypes.

Although they are less common than single nucleotide variants, small deletions or insertions and large gene rearrangements contribute significantly to the severity of the disease. Castaman et al[19] reported four novel null mutations including frameshift deletions and insertions, which caused severe disease in Albanian patients. Nair et al[17] identified 5 deletions and 3 insertions, including cases with double mutations, demonstrating that composite defects can occur and complicate genotype-phenotype interpretation. Reitter et al[18] described double mutations in exon 14 (missense and small deletion) in a family with severe haemophilia A, highlighting the potential for compound effects on FVIII function and inhibitor risk. Large deletions are typically rare but highly consequential, since they are associated with severe deficiency and a high incidence of inhibitors[21] (Table 2).

Table 2 Mutational spectrum of hemophilia A (F8 variants) and associated phenotypes.

Mutation type	Approximate distribution	Associated severity	Molecular mechanism
Inv22[15-20]	40%-50% of severe HA globally; lower in some regions (e.g., 10.5% in Albania, approximately 30% in parts of India/Asia)	Severe	Homologous recombination between int22h-1 and extragenic int22h-2/int22h-3 repeats - disrupted F8 transcription
Inv1[15-17]	2%-5% of severe HA	Severe	Homologous recombination within intron 1
Nonsense mutations[14]	Common among severe, inversion-negative HA	Severe	Premature stop codons - truncated nonfunctional FVIII; often triggers NMD
Frameshift mutations (small insertions/deletions)[17-19]	Frequent in severe HA; multiple novel variants identified in several populations	Severe	Reading-frame disruption - premature truncation
Canonical splice-site mutations[14,17]	Common among severe phenotypes	Severe	Aberrant splicing - exon skipping or truncation
Missense mutations[16,17,22]	Predominant in mild and moderate A; smaller contribution to severe HA (especially at conserved residues)	Mild-moderate; occasionally severe	Residue substitution affects FVIII structure, stability, or cofactor function
Small deletions/insertions (non-frameshift)[22]	Less common but clinically significant	Mild-severe depending on domain affected	Disruption of local protein domains without full truncation
Large deletions/multiexon deletions[21]	Rare	Severe	Loss of entire domains - absent FVIII
Composite (double) mutations[18]	Rare	Severe	Combined effects of two pathogenic variants

HA: Haemophilia A; Inv22: Intron 22 inversion; Inv1: Intron 1 inversion; NMD: Nonsense mutational deletions.

Mutational spectrum in haemophilia B

Haemophilia B results from a highly heterogeneous collection of pathogenic variants in the F9 gene, which spans approximately 34 kb and comprises eight exons. It encodes a modular vitamin-K-dependent serine protease, known as F9. More than 1600 unique variants have now been documented. These affect approximately 88% of FIX residues[23]. Unlike haemophilia A, haemophilia B does not exhibit a dominant recurrent mutation class such as the Inv22.

Missense variants constitute the largest proportion of pathogenic changes in F9 across global cohorts. Structural analyses show that they affect all FIX domains, including the Gla domain, EGF1 and EGF2 modules, the activation peptide, and the serine protease domain[24,25]. In the Italian cohort studied by Bicocchi et al[24], 21 different missense mutations were identified. Each was predicted -through crystallography-informed modeling- to disrupt FIX folding, ligand binding, γ-carboxylation, or catalytic activity. This illustrates the structural sensitivity of FIX, particularly at conserved residues around calcium-binding sites, the activation peptide, and catalytic triad. Several studies underscore that missense mutations disproportionately underlie mild and moderate haemophilia B, thus reflecting residual FIX protein secretion or partial enzymatic function[26]. Yet missense changes can also cause a severe form of the disease when they affect structurally constrained regions or impair post-translational modifications[25]. The updated FIX variant database lists over 945 unique point mutations, most of which only cause ‘small structural perturbations’, consistent with a predominant type II qualitative deficiency[23]. A notable category includes synonymous pathogenic variants. Although they were historically considered benign, several synonymous F9 mutations, including c.459G>A (Val107Val), cause haemophilia B by disrupting the mRNA structure, slowing the translation, and altering the protein folding[27].

Truncating mutations, nonsense variants, frameshift insertions or deletions, and essential splice-site changes, are strongly associated with severe haemophilia B, reflecting total absence of a functional FIX structure. Many truncating variants lead to nonsense-mediated mRNA decay or secretion of unstable protein products[25]. Splice-site alterations are relatively common, given the compact exon-intron structure of F9. Aberrant splicing disrupts domain integrity, as it often produces truncated proteins or impairs signal peptide and propeptide removal, γ-carboxylation, or EGF domain size[25]. Population studies show that truncating mutations form a larger proportion of severe phenotypes compared with missense variants[28].

Large deletions represent only 1%-3% of all F9 mutations, but they are clinically significant because they are often associated with: (1) Severe FIX deficiency; (2) High inhibitor risk; and (3) Extra-haematological features when adjacent genes are co-deleted. Recent genomic analyses have identified various deletions, ranging from single exons to multi-megabase events involving contiguous gene regions. In the most detailed study to-date, Radic et al[29] described 3.9-4.3 Mb deletions that span the entire F9 gene but also neighboring genes SOX3 and MAGEC2, eventually leading to a syndrome of severe haemophilia B combined with severe childhood obesity, global developmental delay, and pituitary hypothyroidism, and this is the first report of such a contiguous gene deletion syndrome[29]. Breakpoint analyses indicated that multiple DNA repair pathways often contribute to these large deletions, including Alu/Alu recombination, non-homologous end joining, and microhomology-mediated break-induced replication[29] (Table 3).

Table 3 Mutational spectrum of haemophilia B (F9 variants) and structural consequences.

Mutation type	Approximate distribution	Associated severity	Structural mechanism
Missense variants[11,23-25]	Approximately 68% of all F9 variants; the dominant mutation class[11]	Mild-moderate most commonly; sometimes severe when essential residues affected	Alter FIX folding, γ-carboxylation, calcium binding, catalytic triad stability, or activation peptide processing
Nonsense mutations[25,28]	Less common than missense; significant proportion in severe HB	Severe	Premature truncation - absent or unstable FIX protein; NMD frequently triggered
Frameshift variants (insertions/deletions)[25]	Relatively uncommon but clinically important	Severe	Frameshift - truncated nonfunctional protein
Canonical splice-site mutations[25,28]	Common among severe HB due to compact exon-domain architecture	Severe	Aberrant splicing - exon skipping, truncated proteins, or defective post-translational processing
Promoter mutations (including HB Leyden)[11]	Rare overall	Childhood severe - spontaneous improvement after puberty	Androgen-responsive elements regulate transcription; puberty FIX expression
Synonymous pathogenic variants[27]	Rare but increasingly recognized	Mild-severe depending on impact on mRNA	Alter mRNA structure, translation rate, and co-translational folding (e.g., p.Val107Val)
Large deletions (partial or whole-gene)[29]	1%-3% of F9 mutations	Severe	Loss of entire exons or full gene; often destabilizes neighboring genomic regions
Contiguous gene deletion syndromes[29]	Extremely rare	Severe haemophilia + syndromic features	Deletion of F9 plus adjacent genes - multi-system phenotype

HB: Haemophilia B; NMD: Nonsense mutational deletions.

THERAPEUTIC ADVANCES IN FACTOR AND NON-FACTOR REPLACEMENTS

General principles of treatment

The management of haemophilia has evolved dramatically over the past decades. Contemporary treatment is grounded on the principles of bleed prevention, joint protection, patient-centered care, along with the pursuit of near-normal haemostasis across the lifespan. Foundational guidance from the World Federation of Hemophilia emphasizes the need for comprehensive and multidisciplinary care, access to effective haemostatic products, genetic assessment and lifelong prophylaxis as the cornerstone of modern management[30]. Historically, patients with haemophilia were provided with on-demand factor replacement, only after bleeding episodes. This approach failed to prevent chronic haemarthrosis and led to cumulative joint damage, disability, and reduced quality of life[31].

The past decade has ushered in transformative non-factor replacement therapies, which rebalance coagulation rather than supplying the missing factor. Emicizumab has already become a routine prophylactic agent in haemophilia A and, indeed, exhibits a favorable safety profile in both clinical trials and real-world use[32].

Extended half-life factors

Conventional treatment relies on intravenous infusions of FVIII for haemophilia A or FIX for haemophilia B. Standard half-life (SHL) recombinant and plasma-derived products effectively treat bleeds and, if used as prohylaxis, reduce their frequency[33]. The introduction of extended half-life (EHL) factor concentrates represents a major achievement in the treatment of haemophilia. Strategies including Fc fusion, albumin fusion, and PEGylation increase the circulating half-life of FVIII or FIX but also reduce the frequency of the infusions and improve the patients’ adherence. FIX EHL molecules, in particular, can achieve 3-5-fold longer half-life and thus enable dosing every 7-14 days[34].

The phase 3A-LONG study demonstrated a 1.5-fold longer half-life than SHL FVIII, accompanied by low median annualized bleeding rates (ABRs) of 1.6 with individualized prophylaxis and 3.6 with weekly dosing[35]. Approximately one-third of the participants achieved 5-day dosing intervals, highlighting the potential for highly flexible regimens[35]. Comparative analyses confirm that while there are small absolute differences in clinical efficacy among EHL FVIII molecules, product-specific differences exist in dosing flexibility, pharmacokinetics, and factor consumption[36]. Most recently, prophylaxis with efanesoctocog alfa provided ultra-long FVIII concentrates, revealed effective bleeding prevention and a normal or near-normal factor activity - in a single weekly dose - both in paediatric and in adult patients[37,38].

Despite the emergence of non-factor therapies, EHL factors remain substantial to haemophilia care. More specifically, they allow: (1) Precise PK-tailored dosing; (2) Ongoing use in patients that need on-demand factor correction (such as trauma, surgery); and (3) Treatment of breakthrough bleeding in patients on non-factor prophylaxis. Moreover, they remain essential in the treatment of haemophilia B, where non-factor options are currently limited[39,40].

The clinical utility of genotyping lies in its ability to predict the risk of inhibitor development[1]. Null mutations (large deletions, nonsense mutations, and Inv22s) that result in a total absence of endogenous FVIII or FIX protein carry the highest immunogenic risk, as the immune system perceives infused factor as a foreign antigen[1]. Conversely, missense mutations that lead to the secretion of a non-functional but structurally similar protein often correlate with lower inhibitor rates, as the patient maintains some degree of immune tolerance to the factor molecule[1].

Factor mimetics (emicizumab)

Emicizumab’s FVIII-mimetic activity arises from its bispecific antigen-binding domains, which are engineered to approximate the spatial configuration of FVIIIa on activated platelets[41]. The efficacy of emicizumab is documented by the pivotal HAVEN trials, which demonstrated remarkable bleeding protection in people with hemophilia A both with and without inhibitors. HAVEN 3 and 4 established that emicizumab provides comparable efficacy in patients without inhibitors. HAVEN 4 showed clinically meaningful bleed prevention with once-every-4-weeks dosing, with 56% of patients reporting zero treated bleeds and a treated ABR of 2.4[42]. Long-term follow-up across HAVEN studies reveals the sustainment of low ABRs, high proportions of zero bleeds, and consistent pharmacokinetics across age groups[43].

Emicizumab has fundamentally redefined prophylaxis in haemophilia A due to its subcutaneous administration, the longer dosing intervals, its efficacy despite the patients’ inhibitor status, and the high adherence and quality-of-life improvement that were widely documented across trials and real-world settings[43,44].

THERAPEUTIC ADVANCES IN GENE THERAPY

Gene therapy for haemophilia extends fundamental principles of in vivo gene addition, whereby a functional copy of the F8 or F9 transgene is delivered to the hepatocytes using a recombinant adeno-associated viral (AAV) vector. This process relies on episomal persistence of vector genomes, hepatocyte-restricted promoters, and long-term transcription within the liver[45]. AAV vectors are non-pathogenic, non-integrating viruses that contain a protein capsid (VP1-3). This capsid surrounds an inverted-terminal-repeat cassette, with serotype determining tissue tropism and immunogenicity[46]. After decades of translational research, the field reached some major milestones with the regulatory approval of the first gene therapies for haemophilia B in 2022 and haemophilia A in 2023[47].

Gene therapy for haemophilia B

Haemophilia B has historically been considered the most amenable target for gene therapy due to the relatively small size of the F9 gene, the low FIX levels required for clinical benefit, and the stability of FIX expression once a therapeutic threshold is achieved. Preclinical and early clinical studies established proof-of-concept more than a decade ago and eventually led to the first durable clinical successes.

The landmark trial by Nathwani et al[48] demonstrated that a single peripheral-vein infusion of an AAV8 vector encoding a codon-optimized FIX transgene (scAAV2/8-LP1-hFIXco) resulted in sustained expression of 2%-11% FIX activity, thus allowing the cessation or spacing of prophylaxis in most participants. This trial set the groundwork for subsequent liver-directed AAV approaches by addressing earlier limitations that had been observed with AAV2 vectors, including the immune-mediated clearance of the transduced hepatocytes[48].

Subsequent generations of AAV-based therapies introduced refinements in vector capsid selection, promoter elements, and transgene design. An important breakthrough was the incorporation of the F9-Padua variant, a naturally occurring FIX mutation (R338 L) with 5-10-fold higher specific activity, which enabled therapeutic FIX activity with lower vector doses[49]. A meta-analysis of haemophilia B gene therapy trials showed a 5.64 reduction in ABR and a mean FIX level of 28.7 IU/mL at 12 months, with nearly complete discontinuation of factor prophylaxis in most patients[50].

Gene therapy for haemophilia A

Gene therapy for haemophilia A has seen rapid progress over the past decade, culminating in the first regulatory approvals and establishing AAV-mediated hepatic gene transfer as a viable therapeutic paradigm. The central strategy is the delivery of a B-domain-deleted FVIII transgene to hepatocytes using an AAV vector. This enables the endogenous FVIII production at levels sufficient to convert the severe phenotype to mild or even normal haemostasis[51].

Valoctocogene roxaparvovec (roctavian) represents the most extensively studied type of gene therapy and the first to achieve regulatory approval[52]. It utilizes an AAV5 capsid and a large approximately 5 kb expression cassette that encodes the codon-optimized FVIII-SQ variant under a hepatocyte-specific promoter. Despite exceeding the typical AAV packaging capacity, this vector reproducibly achieves high initial FVIII expression across phase 1, 2 and 3 trials[52]. In phase 3 studies, valoctocogene roxaparvovec increased FVIII activity to a median 11.9%-62.3% at weeks 49-52, reduced treated bleeding episodes and factor use by > 90%, and substantially improved quality of life[53]. Long-term follow-up from the phase 3 GENEr8-1 study provides a five-year perspective on the durability of valoctocogene roxaparvovec. While the therapy maintains superior bleed control compared to prior prophylaxis, with 81.3% of participants remaining off regular factor infusions at fifth year, a significant multi-year decline in transgene expression is evident[54]. Mean FVIII activity transitioned from a year 1 peak of approximately 42 IU/dL to 13.7 IU/dL at year 5, representing an approximate 60% reduction from peak expression[54]. Updated meta-analytic evidence confirms a reduction of 7.58 annualized bleeding events and a 117-infusion decrease per year relative to pre-treatment[50]. Matching-adjusted indirect comparisons further suggest that valoctocogene roxaparvovec may achieve lower ABRs and higher probability of zero bleeds relative to prophylaxis with emicizumab[55].

In addition to the challenges already described, haemophilia A gene therapy faces important clinical, biological, and implementation barriers. FVIII is a difficult protein for the hepatocytes to synthesize, and emerging molecular studies suggest that UPR activation, ER stress, and intracellular FVIII misfolding may contribute to expression decline[52]. Extremely high vector doses are required, often among the highest used in any human gene-therapy indication, consequently raising safety concerns, among which are hepatotoxicity and potential genomic integration events[46] (Table 4).

Table 4 Comparison of approved gene therapies for haemophilia A and B.

Feature	Valoctocogene roxaparvovec (roctavian)[46,50,52,53]	Etranacogene dezaparvovec (hemgenix)[47-50]
Indication	HA	HB
Vector capsid	AAV5	AAV5
Transgene	B-domain-deleted FVIII, codon-optimized FVIII-SQ	FIX-Padua variant (R338 L), approximately 5-10 × higher specific activity
Mechanism of action	Hepatic expression of FVIII-SQ leads to endogenous FVIII production sufficient to convert severe HA to mild/normal range	Hepatic expression of FIX-Padua generates supraphysiologic FIX activity at low vector doses
Peak factor levels	Median FVIII 11.9%-62.3% at weeks 49-52 in phase 3	Approximately 30 IU/mL at 12 months across trials
Long-term factor expression	Decline over time typical: Approximately 50%-60% reduction from peak by 24 months	FIX expression more stable than FVIII; long-term persistence observed
Reduction in ABR	Meta-analysis: -7.58 treated bleeds/year; > 90% reduction in factor use	Meta-analysis: 5.64-fold ABR reduction; near-universal cessation of prophylaxis
Proportion of patients stopping prophylaxis	Majority (> 90%) discontinue FVIII prophylaxis	Majority discontinue FIX prophylaxis; FIX activity sufficient for stable haemostasis
Durability and challenges	Expression decline over several years; FVIII synthesis is hepatocyte-stressful (UPR/ER stress)	Durable expression; lower dose requirement improves safety margin
Key safety issues	Transaminase elevations common; very high vector dose raises hepatotoxicity concerns potential genomic integration events	ALT elevations possible but generally manageable lower vector dose reduces risk of hepatotoxicity
Advantages	First approved gene therapy for HA; major reductions in bleeding and factor use; high initial FVIII expression	High efficiency due to Padua transgene; stable FIX expression; very low dosing requirements

HA: Haemophilia A; HB: Haemophilia B; ALT: Alanine aminotransferase; ABR: Annualized bleeding rate.

A critical appraisal of current AAV-based gene therapy reveals a significant durability paradox between haemophilia A and B. While FIX expression remains relatively stable over five years, FVIII expression exhibits a characteristic multi-phasic decline, often losing 50% of its peak activity within 24 months[46]. Furthermore, the mandatory use of high-dose corticosteroids to manage transaminase elevations highlights a persistent immunogenicity barrier. The long-term safety profile is further clouded by rare but documented concerns regarding AAV genomic integration near oncogenic hotspots, necessitating lifelong surveillance for hepatocellular carcinoma[46].

CURRENT LIMITATIONS AND FUTURE DIRECTIONS

Despite major therapeutic advances including EHL factors, non-factor rebalancing agents such as emicizumab, and the first successful gene-therapy approvals, substantial challenges are still present in the global, molecular, and clinical management of haemophilia. These limitations concern therapeutic durability, safety, access, health-system readiness, and unmet needs in specific populations, such as paediatric patients. As haemophilia care transitions towards precision medicine and long-acting biological therapies, identifying and addressing these gaps is essential in order to realize the promise of functional or curative treatment.

A major barrier to optimal haemophilia outcomes remains the inequitable global distribution of therapies. Although prophylaxis is recognized as the standard of care, most of the world’s haemophilia population -particularly in low- and middle-income countries- still lacks regular access even to SHL factor concentrates, let alone EHL molecules, emicizumab, or gene therapy[56]. Gene therapies introduce a new layer of economic challenge. Their extremely high upfront cost requires novel reimbursement models, including outcomes-based payments and long-term annuity contracts, to allow sustainable implementation[57]. Clinical challenges persist, even with the modern prophylaxis. Some patients continue to experience breakthrough bleeding, suboptimal joint outcomes, or poor adherence due to treatment burden -especially with intravenous therapies such as FVIII and FIX concentrates[58].

The most significant limitation of current haemophilia gene therapy is the variable and potentially declining long-term transgene expression. FIX expression is comparatively stable, but FVIII expression often shows a characteristic downward trajectory over time, with approximately 50%-60% decline from peak by 24 months across trials[50]. Liver safety remains a central concern. Across trials, transaminase elevation is common and may compromise transgene expression if not promptly treated with corticosteroids[47]. Moreover, accumulated data from AAV therapeutics outside haemophilia, including metabolic disorders, have documented rare fatal hepatotoxicity at high vector doses, underlining the need for cautious dosing, vigilant monitoring, and improved vector engineering[46]. Systematic reviews highlight that no randomized controlled trials comparing gene therapy to standard prophylaxis exist. All current evidence derives from single-arm cohorts or indirect comparisons[58].

Improvements in capsid engineering, such as high-tropism, low-immunogenicity variants, modified promoters, and reduced CpG content, aim to enhance expression, reduce liver toxicity, and broaden eligibility for AAV therapy[47]. A shift towards non-viral platforms offers the possibility of repeated dosing, improved safety, and paediatric use. Lipid-nanoparticle-delivered mRNA, synthetic episomal DNA vectors, and hybrid viral or non-viral systems are all under active development[46]. Next-generation gene editing will form the fundamental target for haemophilia patients by producing sustained and efficient FVIII and FIX, as it was proved by laboratory investigations[59,60]. As emphasized in contemporary reviews, the future of haemophilia management will increasingly rely on personalized treatment strategies, integrating genomic data, pharmacokinetic profiling, patient-reported outcomes, and activity-based goals[56].

CONCLUSION

Significant progress in genetics, non-factor therapies, and gene therapy has transformed haemophilia management, enabling more reliable bleed prevention and improved quality of life. Molecular insights now support precise diagnosis and clearer genotype-phenotype correlations, while agents such as EHL factors and emicizumab have reduced treatment burden. Gene therapy offers the first sustained endogenous FVIII and FIX expression, marking a major step toward functional cure. However, long-term durability, safety concerns, and limited global access remain key challenges. Continued advances in vector design, emerging non-viral platforms, and efforts to reduce treatment inequities will shape the next phase of haemophilia care.

ACKNOWLEDGEMENTS

During the preparation of this work the authors used Google Gemini 3.0 PRO in order to perform language editing. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.