Congenital Hemorrhagic Disorders: New Insights into the Pathophysiology and Treatment of Hemophilia

David Ginsburg, M.D.^*

Classic hemophilia is due to mutations in either the factor VIII (FVIII) or factor IX (FIX) genes, classified as hemophilia A and hemophilia B, respectively. Both genes are located on the X chromosome, leading to the classic X-linked inheritance of these disorders that has been recognized since biblical times. Defects in a number of other coagulation factors cause very similar clinical phenotypes, though these disorders are considerably less common than the hemophilias and readily distinguished by laboratory testing for specific coagulation factors. With the advent of modern molecular genetics, the genes for all of the well-characterized clotting factors have been cloned and sequenced, and a wide spectrum of mutations in the FVIII and FIX genes have been identified. This section will review the current state of knowledge concerning the molecular basis of hemophilia A and hemophilia B and potential clinical applications of DNA diagnosis, including prenatal testing. Two other recently identified genetic disorders resulting in markedly reduced FVIII levels will also be discussed: von Willebrand disease (VWD) type 2N and combined deficiency of FV and FVIII.

Hemophilia A is due to absent or decreased FVIII procoagulant function, resulting from mutations in the FVIII gene. A number of other congenital coagulation factor deficiencies can present with a similar picture of mild to severe clinical bleeding. However, the diagnosis of hemophilia A is generally readily established by the identification of an isolated deficiency of plasma FVIII activity.^1,2  An X-linked pattern of inheritance is also helpful, although this does not distinguish hemophilia A from hemophilia B (factor IX deficiency). In addition, approximately one-third of patients represent new mutations and have a negative family history (see below).

The overall prevalence of hemophilia is usually estimated at between 1:5,000 and 1:10,000 males. A recent detailed epidemiologic survey yielded a projected total of 13,320 cases of hemophilia A and 3,640 cases of hemophilia B in the U.S. population. This corresponds to a prevalence of approximately 1:10,000 males for hemophilia A and 1:35,000 for hemophilia B. The combined incidence for both hemophilias was estimated to be approximately 1:5,000 live male births.³ 

The severity of bleeding associated with hemophilia A can be accurately predicted by the level of residual FVIII or FIX activity in plasma. Factor levels of < 1% of normal are associated with severe hemorrhagic symptoms, 1-5% levels with moderate hemophilia, and levels of 5-25% with only mild disease. Approximately 70% of hemophilics are classified as severe, though this number may represent an overestimate since severe hemophilics are more likely to seek medical care. The epidemiologic survey described above identified approximately 43% of hemophilics as severe, with 26% classified as moderate and 31% as mild.³  Treatment of hemophilia A with FVIII replacement, reviewed in another section of this update, has produced dramatic improvement in the life expectancy of hemophilics. In the early 1900s, median life expectance was only 11.3 years whereas it is currently estimated to be between 60-70 years.⁴ 

The FVIII gene spans 186 kb dispersed across 26 exons near the tip of the long arm of the X chromosome (Xq28). The FVIII mRNA is approximately 9 kb in length and encodes an approximately 300 kD protein.⁵  The primary source of FVIII production in vivo is still unknown, although significant production can occur in the liver, based on the correction of hemophilia A by liver transplantation. Gene expression has also been identified in peripheral blood cells, as well as some lymphocyte cell lines.

The FVIII protein circulates in plasma at very low concentrations (approximately 100 mg/ml) where it serves as a critical co-factor for FIX in the proteolytic activation of factor X to its active form (FXa). FVIII is an intrinsically unstable protein, requiring strong electrostatic interaction with van Willebrand factor (VWF) for stability in plasma. The two proteins thus circulate together as a tight complex in plasma.⁵  The extreme instability of FVIII in the absence of VWF accounts for the markedly reduced FVIII levels observed in severe VWD patients and patients with homozygous type 2N VWD (see below).

The Haldane hypothesis predicts that one-third of all patients with an X-linked lethal disorder should represent new mutations. This results from the fact that one-third of all X chromosomes reside in males and two-thirds in females. Prior to the modern medical era, the one-third of hemophilia chromosomes present in males would be lost from the population with the early death of severely affected patients. Because the frequency of the disease is in equilibrium, these lost hemophilia alleles could be assumed to be replaced by an equal number of new mutations. This general principle applies to many other X-linked diseases such as muscular dystrophy.

As noted above, the frequency of hemophilia A is generally estimated at between 1:10,000 and 1:5,000 males. Based on this gene frequency, approximately 1:25,000,000 to 1:100,000,000 females would be expected to be homozygous for a factor VIII deficient allele. Consistent with these predictions, classic hemophilia A in females has only very rarely been reported and other possible explanations should be considered whenever a female with this phenotype is encountered. Extreme skewing of X chromosome inactivation can result in unusually low factor VIII levels in female hemophilia A carriers. Though this may produce reductions of factor VIII levels into the mild or even moderate hemophilia A range, levels ≤ 1% and the associated severe bleeding are extremely unlikely. The exception is the rare individuals who carries a hemophilia A mutation associated with an X chromosome:autosome translocation or other cytogenic abnormality, which may result in exclusive inactivation of the normal X-chromosome. If the rearranged X-chromosome also carries a hemophilia A mutation, then this female could present as a classic hemophilic.⁶  Finally, the diagnosis of homozygous type 2N VWD should also be considered in any female with very low factor VIII levels (see below).

With the high frequency of new mutations predicted by the Haldane hypothesis, one would generally expect to encounter different molecular defects in any two unrelated hemophilia A patients. This is indeed the case for point mutations identified within the gene by DNA sequence analysis (see below). However, a unique rearrangement within the FVIII gene has recently been identified as a common, recurrent mechanism for hemophilia A.^7,8  This rearrangement is an inversion within the FVIII gene that is the result of unequal crossing over between a duplicated sequence within intron 22, termed gene “A,” and two highly homologous blocks of sequence located downstream of the FVIII gene, toward the telomere of the X chromosome long arm. The rearranged gene is disrupted at intron 22. The resulting truncated protein product is presumably unstable and does not accumulate to significant levels. Thus, this common FVIII gene inversion is associated with undetectable levels of FVIII (< 1%) and a typical severe hemophilia A phenotype. It is thought that the specific configuration of the FVIII gene and the repeated gene A elements, as well as their location on the long arm of the X chromosome, make this region particularly prone to rearrangement, leading to the high frequency of this recurrent mutational event, which is now known to account for approximately 45% of all severe hemophilia A patients.⁸ 

Another unique feature of the recurrent FVIII intron 22 gene inversion is the observation that the mutation almost always arises during a male meiosis.^8,9  This is presumably due to the large region of nonhomology between the X and Y chromosome during meiotic pairing, favoring a misalignment between the gene A repeated sequences and an illegitimate recombinant event, resulting in the gene inversion. The presence of a second X chromosome with full pairing in this region inhibits this event in a female meiosis. This observation has important clinical implications. For an apparently new mutation patient in whom a gene inversion is identified, the mother can generally be assumed to be a carrier, with the recombination event often identified in the maternal grandfather's allele.

The remaining ∼55% of severe hemophilia patients who do not carry the intron 22 inversion can generally be shown to have a more conventional molecular defect in their FVIII gene. Approximately 5% of patients have deletions removing varying sized segments of the FVIII gene. A number of small insertions and deletions have also been defined. The remaining large group of patients (∼50%) are generally found to have specific point mutations in exons or at splice junctions within the FVIII gene.¹⁰ 

In addition, nearly all patients with mild or moderate hemophilia A (in which some residual level of FVIII activity can be demonstrated) can be shown to have a point mutation within the FVIII coding sequence, resulting in a single amino acid substitution. Depending on the location and nature of the amino acid substitution, a range of residual FVIII activities are observed, accounting for the variation in severity among mild and moderate patients. Though the large size and complexity of the FVIII protein has been a major obstacle to direct structural analysis, an x-ray crystallographic structure of the FVIII C2 domain was recently reported,¹¹  providing new insight into the molecular pathogenesis of hemophilia A resulting from mutations in this region of the molecule.

As has been observed in a number of other genes, mutations resulting from C (cytosine) to T (thymidine) transitions at CpG dinucleotides are particularly common, accounting for approximately one-quarter of single-based substitutions.¹⁰  This hypermutability appears to result from the frequent methylation of Cs at CpG dinucleotides. Methylcytosine can spontaneously deaminate to form uracil, which will be converted by the cells DNA repair machinery to thymidine.

Because any two unrelated hemophilia A patients can generally be assumed to have different mutations, molecular studies of this population have provided important insights into the general mechanisms of human mutation. In addition to the large number of typical point mutations and deletions and insertions described above, as well as the unique but recurrent factor VIII gene inversion, a number of other unusual mechanisms for mutation have been identified in hemophilia A patients, including the novel insertion of a human transposable sequence termed a LINE element.¹² 

A database of factor VIII gene mutations has been maintained by a consortium of investigators in the field with periodic published updates, the most recent in 1994.¹⁰  More up-to-date information is also available at http://europium.csc.mrc.ac.uk/usr/WWW/WebPages/main.dir/main.htm. The database currently catalogues the results of analyses for over 1,000 DNA samples from hemophilia A patients. As of May 1999, 309 single-based substitutions had been described with 264 (85%) leading to a single amino acid substitutions (missense mutations) and 45 (15%) to premature stop codons (nonsense mutations). An additional 38 mutations may lead to aberrant mRNA splicing. Ninety-two gene deletions have been reported, ranging from 1 kb to 210 kb in size and overall accounting for about 5% of hemophilia A patients. There are also 77 reports of small (< 100 base pair) deletions ranging from 1-86 base pair in size. Most of these small deletions result in a frameshift and loss of FVIII expressions. Finally, 28 mutations represent insertions of from 1 base pair to 3.8 kb in size, again usually leading to a frameshift and severe hemophilia A.

As noted above, the clinical severity of the hemophilia A phenotype correlates very closely with the amount of residual factor VIII activity. However, given the complexity of hemostasis, one could easily imagine that variation at other coagulation factor loci could impact on hemophilia A severity.

Consistent with this notion, a few cases have now been reported in which patients with identical FVIII gene missense mutations exhibit significantly different degrees of clinical severity. For example, an arginine to cysteine substitution at amino acid 1689 has been reported in several patients, some exhibiting mild hemophilia A and others a moderately severe phenotype. A similar variation in phenotype has been noted among patients carrying a glutamine to arginine substitution at position 2209.¹⁰  Since these differences in phenotype cannot be readily explained simply on the basis of the FVIII gene defect, the influence of other modifying genetic factors should be considered.

A recent report suggests that the common FV Leiden thrombophilia mutation may be such a modifying factor. Two pairs of patients, one with the Arg1689Cys mutation and one with the Gln2209Arg mutation were reported in which one member of each pair carried the FV Leiden mutation and the other the normal FV sequence. In both cases, the patient also inheriting the FV Leiden defect exhibited a significantly milder clinical disorder.¹³  An in vitro analysis of clotting factor function also supports the hypothesis that FV Leiden can increase net thrombin formation in hemophilia A patients.¹⁴  Given the high frequency of the FV Leiden mutation in the general population (approximately 5%), this could be an important factor in a significant subset of patients^6,15  This potential interaction between FV Leiden and hemophilia A has been confirmed in one subsequent study¹⁶  but not another.¹⁷  Other common thrombophilia mutations, such as the heat-labile methylenetetrahydrofolate reductase (MTHFR) mutation and the recently identified prothrombin gene polymorphism,^18,19  have not yet been evaluated as potential modifiers of hemophilia A.

The remarkable advances in the molecular understanding of hemophilia over the past 10 years have made this disorder a prime target for gene therapy approaches, a topic discussed in another section of this update. These advances have also led to the production of recombinant FVIII products for the treatment of hemophilia patients, also reviewed elsewhere in this update.

Another clinical arena in which molecular advances have had an important impact on practice is in the area of prenatal diagnosis. As discussed above, a precise genetic diagnosis does not generally alter the clinical management of a known patient with hemophilia A, since the severity of the bleeding phenotype is primarily determined by the residual level of FVIII activity, as measured by standard clotting assays. Though the development of inhibitors to FVIII is a major problem in clinical management (see later section), this event cannot be readily predicted from knowledge of the molecular defect. In fact, recent studies have demonstrated that the large cohort of patients with the FVIII gene inversion, all carrying the identical molecular defect, show a similar frequency of inhibitor development (approximately 10-20%) as an unselected group of severe hemophilia A patients.⁸ 

Clinical genetic testing for the common intron 22 inversion is now offered by several commercial DNA laboratories. As noted above, this analysis will provide the precise genetic diagnosis in approximately 45% of severe hemophilia A patients. This test should not be ordered in patients with moderate or mild hemophilia A, since the presence of detectable levels of FVIII activity in these patients excludes a possible DNA gene inversion diagnosis.

The molecular defects in the remaining 55% of severe hemophilia A patients not detected by the intron 22 inversion test, as well as all mild and moderate hemophilia A patients, can be detected in most cases by efficient screening of all 26 FVIII exons and splice junctions.²⁰  Such direct DNA sequence screening has recently become available through several commercial DNA diagnostic laboratories. In rare patients in whom the precise mutation cannot be identified, accurate genetic diagnosis often still can be achieved by a linkage approach. In these studies, FVIII alleles of an at-risk female potential hemophilia A carrier are distinguished on the basis of polymorphic DNA sequence differences between the two copies of the gene. A number of intragenic polymorphisms are available that should lead to an informative analysis in over 90% of families. If the two X chromosomes of the putative carrier can be distinguished, analysis of an affected male will establish phase, that is identify which of the two X chromosomes carries the hemophilia A gene mutation, without actually knowing the precise mutation itself. Analysis of any subsequent at-risk pregnancy can then identify which of the two maternal FVIII gene copies is present and thus establish whether the fetus is normal or has inherited hemophilia A. With appropriate informative markers, this analysis is extremely accurate, though the possibility of a recombination between the informative marker and the actual mutation must always be kept in mind.

Direct mutation testing or DNA linkage analysis can be performed on a chorionic villus biopsy sample obtained as early as 10 weeks of pregnancy. Prior to the availability of DNA testing, the only option for prenatal diagnosis was direct fetal cord blood sampling and testing for FVIII clotting activity. Although accurate, this approach can only be performed quite late in pregnancy and carries significant risk of fetal loss due to the procedure itself. However, the latter might still be employed for diagnostic confirmation, or in those rare cases for which neither the precise mutation nor informative DNA markers are available.

Implicit in prenatal diagnostic testing by linkage analysis is the assumption that the mother is a hemophilia A carrier. This can be difficult to establish definitively for families in which the proband may represent a new mutation. Genetic probability analyses can often provide more accurate risk estimate in such cases and genetic consultation should be considered.

VWD is an exceedingly common inherited bleeding disorder with prevalence estimated to be as high as 1% in several populations.^21,22  As a result of the dependence of FVIII stability in the circulation on its association with VWF (see above) the reduced levels of VWF in most VWD patients are also associated with a reduction in plasma FVIII activity. Indeed, parallel reduction of VWF (measured as either ristocetin co-factor activity or VWF antigen (“FVIII related antigen”) and FVIII activity to the range of 20-50% of normal is the hallmark of type 1 VWD.^23,24  The type 2N variant of VWD, first described in the early 1990s, can lead to isolated deficiency of FVIII activity in the presence of otherwise normal VWF. Type 2N VWD results from mutations within the FVIII binding domain of VWF, leading to selective loss of this important function.²⁵  Type 2N VWF otherwise exhibits normal platelet adhesive function and bleeding in these patients appears to result solely from the decreased FVIII procoagulant activity. The mutations in type 2N VWD are clustered at the N-terminus of the mature VWF subunit,²⁶  in a region shown by structure function studies to contain the FVIII-binding domain. The most common mutation, arginine 91 to glutamine, causes a partial decrease in FVIII binding. Patients homozygous for this mutation generally have FVIII levels in the range of 10-25%. Other substitutions have more marked effects on FVIII binding.²⁷  Very rarely, mutations in this region may reduce FVIII levels to the range of a moderate or severe hemophilia patient. Thus, type 2N VWD generally need only be considered in the differential diagnosis in a mild or moderate hemophilia A patient and is unlikely to be the explanation in a patient with undetectable FVIII levels. The arginine 91 to glutamine mutation may be as frequent as 1% in some populations. Co-inheritance of this or one of the other type 2N mutations with a VWF type 1 or null allele is another mechanism for reduced FVIII levels.²⁸  Type 2N VWD should be particularly considered in families with unusual patterns of hemophilia inheritance, including affected females.

DNA testing for a limited set of type 2N VWF gene mutations has recently become available at several commercial DNA diagnositc laboratories. Testing of patient plasma VWF for reduced FVIII-binding activity is also offered as a useful screening test through specialized reference clotting laboratories.

First recognized over 40 years ago, combined deficiency of FV and FVIII is associated with simultaneous reductions in activity for both factors to the range of 10-15% of normal. This disorder was recently shown to be due to mutations in ERGIC-53, a protein that appears to be required for the efficient transport of FV and FVIII from the endoplasmic reticulum to the Golgi.^29,30  Inheritance is autosomal recessive and associated with complete loss of ERGIC-53 function. Recently, a subset (approximately 25% of combined FV and FVIII deficiency patients has been shown to have mutations in another gene distinct from ERGIC-53, though the responsible gene remains to be identified.^31,32  Well over 100 families with combined FV and FVIII deficiency have now been reported, most clustered in southern European countries or the Middle East. Diagnosis of this disorder is currently based solely on assays of plasma FV and FVIII activity and family studies, as DNA analysis is only available on a research basis.

The clinical presentation of hemophilia B is nearly indistinguishable from that of hemophilia A, though the two disorders can easily be separated on the basis of routine laboratory testing and the measurement of FVIII and FIX activity.^2,6  Both the FVIII and FIX genes are located on the long arm of the X chromosome, separated by approximately 10 centimorgans. As noted above, hemophilia B accounts for approximately 20-25% of all hemophilia cases.³  As for hemophilia A, the severity of disease is closely correlated with the residual level of FIX activity.

The FIX gene is 34 kb in length and contains 8 exons encoding a 461 amino acid precursor protein. FIX is a member of the serine protease gene family and, together with its nonenzymatic cofactor VIII, forms the “Xase” complex that cleaves the serine protease zymogen FX to generate the active enzyme FXa. Along with factors II, VII, X, protein C, and protein S, FIX is dependent on post-translational processing by γ-carboxylase for full functional activity. γ-carboxylase converts selected glutamic acids in the propeptide domain to γ-carboxylglutamic acid, or GLA residues. The FIX gene is expressed primarily in the liver, which also contains high levels of γ-carboxylase activity, leading to efficient production of functional FIX that is subsequently secreted into plasma and circulates at a concentration of approximately 10 μg/ml, nearly 100-fold higher than the plasma concentration of FVIII. The transcriptional regulation of the FIX gene has been a subject on intense study, due in large part to the unique series of mutations giving rise to a variant of hemophilia B referred to as FIX Leyden. These individuals have very low levels of FIX and severe hemophilia as children, but show dramatic improvement at puberty with a rise of FIX levels to near normal range.³³  A number of mutations in the promoter region of the FIX gene have been identified in these patients contributing to our understanding of FIX gene regulation.³⁴ 

The study of FIX gene mutations has provided valuable information about the rates and types of mutations that occur in human populations. Approximately 1/3 of the point mutations seen in hemophilia B have been shown to occur at CpG dinucleotides.³⁵  As in hemophilia A, approximately one-third of mutations in hemophilia B appears to have arisen de novo, consistent with the Haldane hypothesis. Recent analysis of a large cohort of hemophilia B patients it the United Kingdom identified an overall human mutation rate of 2.14 × 10^-8 per base per generation or 128 mutations per human zygote, with an estimated 1% of these (approximately 1.3 per zygote) being detrimental.³⁶  A database of reported FIX gene mutations is maintained at the following website: http:// www.umds.ac.uk/molgen/haemBdatabase.html

The most recent summary of these data (Version 9) reveals 1,918 patient entries, including a number of gene deletions of varying sizes and over 1,500 point mutations. There appears to be an increase in the incidence of inhibitor antibody development in patients with large gene deletions.³⁷  As for hemophilia A, direct DNA mutation testing is available through several clinical DNA diagnostic laboratories, as well as linkage analysis for those cases in which the responsible mutation is not readily identified. DNA testing can be used for prenatal diagnosis or to definitively establish carrier status in at-risk females.

Remarkable progress over the past 15 years has provided a wealth of information about the molecular basis for hemophilia A and hemophilia B. Though modern molecular diagnosis is only beginning to be offered to patients in the clinical setting, this new technology has the potential for considerable value in enhancing patient care. With the current “state-of-the-art” genetic testing, precise prenatal diagnosis should be possible as early as 10 weeks of gestation for nearly all pregnancies known to be at risk. Taken together with recent progress in gene therapy and the production of recombinant products for the treatment of hemophilia (see subsequent sections of this update), the translation of these molecular advances from the “bench to the bedside” should serve as a useful paradigm for similar approaches to many other genetic disorders.

Jeanne M. Lusher, M.D.^*

Despite greatly improved methods for donor screening and viral attenuation of plasma-derived clotting factor concentration over the past decade, sporadic reports of transmission of blood-borne viruses such as hepatitis A and human parvovirus B 19 have resulted in continued concerns about safely.^1,2,3  Additionally, periodic withdrawals of plasma-derived products in the US through 1998 because of donors later found to have Creutzfeldt-Jakob Disease (CJD),^4,5  and periodic shut-downs of concentrate manufacturing facilities for months at a time, have led to supply concerns.

Fortunately, recombinant FVIII (rFVIII) and FIX (rFIX) concentrates are now available, having been licensed by the U.S. Food and Drug Administration in 1992 (Baxter's rFVIII, Recombinate®), 1993 (Bayer's rFVIII, Kogenate®), and 1997 (Genetics Institute's rFIX, (Bene-FIX®). In the spring of 2000, Genetics Institute's B-domain deleted rFVIII, ReFacto®, and Bayer's Kogenate® FS were licensed; however, as of August 30, neither had yet become commercially available in the US. Largely because of their increased margin of viral safety, the use of these recombinant FVIII and FIX concentrates has steadily increased. Surveys conducted by the Marketing Research Bureau indicated that rFVIII concentration accounted for ∼78% of FVIII used in the U.S.⁶  in 1999, while rFIX accounted for ∼80% of FIX used.⁶ 

Prior to licensure of each of the rFVIII products, extensive in vitro and preclinical studies in animals were conducted, followed by pre-licensure clinical trials in persons with hemophilia A. The pre-licensure clinical trials with the two full-length rFVIII preparations began in 1987 and 1988, and demonstrated the safety and efficacy of these products in a wide variety of settings (surgical coverage and other in-hospital use, home treatment, prophylaxis, etc.).^7,8,9,10  In prospective studies in previously untreated patients (PUPs), 30-35% of PUPs developed inhibitory antibodies to FVIII after relatively few exposure days (ED) to rFVIII (median 10-12 ED); however, at least a third of these were low-titer inhibitors (< 5 BU), and many of these disappeared spontaneously (i.e., while the child remained on episodic, “on demand” treatment with rFVIII).^11,12  A number of others responded well to immune tolerance induction (ITI) regimens (with rFVIII alone).

These observations are similar to those obtained in prospective studies with plasma-derived products.^13,14  It is now apparent that genetic factors play a major role in FVIII inhibitor development (e.g. the underlying defect in the FVIII gene causing the hemophilia,¹⁵  or racial background, with individuals of African^16,17  or Hispanic¹⁸  background having a higher likelihood of inhibitor formation). Immunologic factors no doubt play a role as well. While there have been two instances in European countries in which a pasteurized or double-virus inactivated plasma-derived FVIII concentrate has resulted in a higher than expected number of inhibitors (in previously treated patients)^19,20  none of the rFVIII or rFIX preparations have been so implicated.

While the two full-length rFVIII preparations have had an excellent record of safety (with 12-13 years of experience in persons with hemophilia A), some were concerned because these recombinant FVIII preparations are stabilized with human serum albumin (which in fact is the major component of these rFVIII concentrates). Thus, scientists working with pharmaceutical companies devised other ways of stabilizing rFVIII. The new formulation of Bayer's rFVIII is stabilized with sucrose rather than albumin.^21,22  This formulation, named Kogenate FS, entered pre-licensure clinical trials in 1996 and was licensed for use in the US in 2000. Studies to date have shown this product to be effective, safe, and no more immunogenetic than other FVIII products.^21,22  A “second generation” rFVIII product, Genetics Institute's B-domain deleted (BDD) rFVIII (ReFacto) entered pre-licensure clinical trials in 1993, and was licensed for use in Europe in 1998 and in the U.S. in March, 2000. The design of BDD rFVIII was based on current knowledge of the molecular structure and function of FVIII. The heavily glycosylated B domain seems to be dispensable for the hemostatic activity of FVIII,²³  and deletion of the B domain results in a molecule which is more easily secreted by Chinese hamster ovary (CHO) cells and is less prone to proteolytic degradation. Additionally, no albumin is needed as a stabilizer. Biochemical characterization studies have demonstrated that BDD rFVIII is essentially identical to plasma-derived FVIII with respect to functional properties,^24,25  von Willebrand factor-binding kinetics²⁶  and pharmacokinetics.²⁷  However, it should be noted that, in patients infused with BDD rFVIII, one-stage APTT-based FVIII assays give lower than expected results (on average, 50% less)²⁸  while two-stage and chromogenic assays give expected (calculated) values. This assay discrepancy has been shown to reflect a difference in phospholipid binding with BDD rFVIII.²⁹  Since almost all clinical laboratories use one-stage APTT-based FVIII assays, one should be aware of this. If one looks at the one-stage FVIII results in a patient receiving BDD rFVIII and calculates subsequent dosing based on this, you will be over treating the patient!

The prelicensure prospective studies with BDD rFVIII in previously treated patients (PTPs), PUPs, and in individuals with hemophilia A undergoing surgical procedures represent the largest prospective trials with a FVIII product. It is noteworthy that this bioengineered, B-domain deleted product, has proven to be highly effective, safe, and no more immunogenic than plasma-derived or full length rFVIII concentrates.³⁰ 

The FIX gene and cDNA were first cloned in 1982,^31,32  and soon thereafter rFIX was expressed from cDNA in CHO cells.³³  Prelicensure clinical trials with Genetics Institute's rFIX (BeneFIX) began in 1995; the product was licensed by the U.S. FDA in February 1997. The CHO cell line used in the manufacture of rFIX is cotransfected with a human recombinant FIX cDNA expression plasmid and a cDNA expression plasmid that encodes an engineered form of the paired amino acid cleaving enzyme (PACE). PACE improves the processing efficiency of profactor IX expressed in CHO cells.³³  The CHO cells are grown in a medium that contains no animal- or human-derived proteins. No albumin is used, and no human plasma, animal plasma or animal-derived protein are used in its manufacture or purification.³⁴  Thus, BeneFIX is virtually risk free in terms of transmission of blood-borne viruses and spongiosiform agents.³²  The product has proven to be effective for treatment and prevention of bleeding, has an excellent track record or safety, and there has been no increased incidence of FIX inhibitor development with its use.

In view of the lower recovery values following infusion of BeneFIX,³⁴  which may be as low as 50% of expected recovery in some recipients, it is recommended that one use a somewhat larger dose of this product than one would of a plasma-derived FIX concentrate. The product package insert recommends the following calculation of dosage: number of FIX units required = body weight (kg) × desired FIX increase (%) × 1.2. However, one should be aware of the wide variations in recovery among individual patients. Infants and young children tend to have the lowest recoveries.³⁵  The difference in recovery has been shown to be due to a simple difference in the post-translational modification of rFIX, namely, the differences in sulfation of tyrosine 155 and phosphorylation of serine 158, residues which appear to play a role in the clearance of FIX.³⁴ 

Recurrent bleeding into joints and soft tissue is the hallmark of severe hemophilia A and B. Episodes of acute hemarthroses usually begin around 1 or 2 years of age³⁶  when the child is learning to walk and has frequent falls and collisions with tables or other hard objects. Until the availability of clotting factor replacement therapy, most developed severe musculoskeletal problems that often led to crippling deformities.^37,38,39,40 

If chronic joint disease is to be prevented in individuals with severe hemophilia, prophylactic infusions of FVIII or FIX must begin at a relatively early age. Persons with mild (baseline FVIII or FIX > 5-30%) or moderate (1-5%) hemophilia rarely have spontaneous bleeding into joints, and have a much lower prevalence of chronic joint disease than do persons with severe hemophilia (those with a baseline FVIII or FIX of <1%).^39,40  Thus most prophylactic regimens are aimed at keeping trough levels of FVIII or FIX over 1%.

Professor I.M. Nilsson and co-workers in Malmö, Sweden, began prophylaxis for boys with severe hemophilia in 1958.³⁸  Over the next three decades, Nilsson et al improved outcomes by increasing the dose of clotting factor, decreasing the dosing intervals, and decreasing the age at which prophylaxis was begun.^39,41,42  Petrini and colleagues in Stockholm also began instituting prophylaxis at an early age and reported excellent results.⁴³  After a thorough review of the Swedish experience with prophylaxis, in early 1994 the United States' National Hemophilia Foundation's Medical and Scientific Advisory Council (MASAC) issued a series of recommendations concerning prophylaxis.⁴⁴  In part, the MASAC recommendations state that (1) in view of the demonstrated benefits of prophylaxis begun at a young age in persons with hemophilia A and B, MASAC recommends that prophylaxis be considered optimal therapy for children with severe hemophilia A and B; (2) prophylactic therapy should begin early (at 1-2 years of age) with the aim of keeping the trough FVIII or FIX level above 1% between doses. This can usually be accomplished by giving 25-40 FVIII units/kg three times per week, or 25-40 FIX units twice weekly; (3) a periodic mechanism should be developed and used to evaluate joint status, to document any complications, to document all costs associated with the child's prophylaxis, and to record any bleeding episodes that occur during prophylaxis. The MASAC recommendations also encourage parents of young children with severe hemophilia to discuss with their health care providers the risks versus benefits of prophylaxis (including potential cost and reimbursement issues, possible venous access problems necessitating the use of a central venous access device, such as a surgically implanted port, and potential complications of such devices).⁴⁴ 

While not all families choose, or can afford, prophylaxis for their child, this approach should be thoroughly discussed with each family, so that they can make an informed decision.^44,45  More and more US children with severe hemophilia are being started and maintained on prophylaxis. It is clear that hemophilic arthropathy can be prevented.^41,45,46  In order for prophylaxis to be successful, however, a major long-term commitment must be made by the child's parents.⁴⁵  It is important to optimize the prophylactic regimen so that cost and clinical outcome are acceptable to the family, the care providers and payors.^46,47,48  A recent analysis of 121 Swedish patients on prophylaxis again demonstrated the importance of starting early (before the age of 3 years); however, the authors note that the frequency of infusions at the start of prophylaxis could be individualized and adjusted according to the bleeding pattern, thus obviating the need for central venous access device in some young children.⁴⁹  However, while most agree that prophylaxis should begin at an early age, the ideal time at which to begin prophylaxis remains a subject of debate.^50,51 

In recent years, radionuclide synoviorthesis has gained in popularity as clinical experience in an increasing number of medical centers has demonstrated its effectiveness (including cost-effectiveness) and safety. While the greatest experience with radionuclide synoviorthesis has been in patients with rheumatoid arthritis, this procedure is now being performed in children (≥ 5 years of age) and adults with hemophilia. Results are much better in those with early chronic synovitis than in those with later stages of chronic synovitis or destructive joint disease.

While several radioactive materials have been used, P32 is the preferred one in the U.S. The fibrosing of synovium by a radioactive material has several advantages over surgical synovectomies. It can be done on an out-patient basis, it requires very little replacement therapy (a therapeutic level of FVIII or FIX for 72 hours), and does not result in restricted range of motion. Results are similar to those reported for surgical synovectomy, with an approximate success rate of 80%.^52,53,54,55  However, some groups prefer arthroscopic synovectomies, either “standard”⁵⁶  or with a Holmium:Yag laser.⁵⁷ 

Noting a lack of long-term follow-up data in hemophilic patients undergoing various types of interventions, in 1999 the National Hemophilia Foundation's MASAC recommended that collaborative efforts be established to collect data regarding interventions to prevent joint damage secondary to hemophilia.⁵⁸ 

Ulla Hedner, M.D.^*

Inhibitors against FVIII and FIX develop in around 15% of patients with severe hemophilia¹  and are directed against the procoagulant part of the FVIII or FIX molecules. Patients with low inhibitor titer, and especially those with a low anamnestic response, can be given high doses of the specific factor concentrate to neutralize the inhibitors and to induce hemostasis. However, for patients with high inhibitor levels or high-responders, other treatments must be used, including procedures to decrease the antibody titers. To minimize the booster effect of high doses of antigen, immunosuppressive treatment may be added. Because such treatment procedures are complicated and associated with a number of potential side effects, they are not often used to treat mild-to-moderate bleeding episodes or to cover elective, less urgent surgery for inhibitor patients. As a result, a great deal of effort has been devoted to finding more convenient treatment modalities and to inducing immunologic tolerance in order to permanently eradicate the inhibitors.

Hemostasis may be achieved in patients with inhibitor titer below 10 BU/ml (Bethesda units) by administration of FVIII or FIX concentrates in sufficiently large doses. It is, however, important to make sure the inhibitors are totally neutralized in both the plasma and extravascular space and that a hemostatic level of 60 U/ml to 100 U/ml of FVIII or FIX is achieved. The dose is calculated using the patient's inhibitor titer, hematocrit, and body weight and is based on the assumption that the extravascular space is equal to the plasma volume and, therefore, has the same inhibitor titer. Following the neutralizing dose, FVIII or FIX are given in concentrations ranging from 60 U/kg to 100 U/kg body weight, followed by dosing at 8 to 12 hour intervals until the bleeding is under control.^2,3 

Following such a treatment low-responding patients may experience only a modest booster effect with slightly increased inhibitor titers. Patients who have previously shown a substantial booster response with markedly increased inhibitor levels following the administration of FVIII or FIX concentrates, are recommended to receive immunosuppressive treatment for example cyclophosphamide as adjunct therapy.²  The combination of high doses of antigen and immunosuppressive therapy was introduced by Green⁴  and was based on observations that antibody production had been suppressed in other situations if antigen and cytotoxic drug were administered concurrently.

Successful treatment of acute bleeding episodes in hemophilia patients with inhibitors has been reported by using the combination of high antigen doses and immunosuppressive treatment.^2,5,6  Following repeated episodes of this type of combined therapy, patients who previously had been high-responders became low-responders, which indicated that partial immunologic tolerance had been achieved.^7,8 

In patients with very high antibody titers initially, it may be impossible to neutralize all intra- and extravascular antibodies and add enough FVIII or FIX to reach a hemostatic plasma level. For these patients, a number of complicated procedures have been used to temporarily lower the inhibitor titer. First, extensive plasmapheresis with the administration of FVIII concentrates and fresh-frozen plasma has been used successfully in a few cases.⁹  Used in combination with immunosuppressive therapy, the booster effect of the FVIII concentrate was diminished.¹⁰  Also, the extracorporeal antibody adsorption may be performed by immunoglobulin adsorption on protein A¹¹  or by specific immunoadsorption onto immobilized FVIII or FIX.¹² 

These rather complicated procedures are difficult to use in the management of acute bleeding episodes because access to sophisticated instrumentation and skilled personnel are required. The patient must be hemodynamically stable and have good venous access. Also, hours or days may be required to achieve an adequate antibody reduction. As a result, these procedures are best suited to the prophylactic removal of antibody prior to an anticipated procedure.¹³ 

Permanent eradication of the FVIII or FIX inhibitors is the ultimate goal in the treatment of inhibitor patients and a number of treatment schedules have been proposed to induce immunologic tolerance. Immunosuppressive treatment alone has had limited success in patients with alloantibodies.¹⁴  However, the combination of high doses of antigen and immunosuppressive treatment (cyclophosphamide) has been demonstrated to minimize the booster effect and, when repeated, to convert high-responding patients into low-responders or to induce at least partial immunologic tolerance.^7,8 

The addition of high-dose intravenous IgG to a treatment including FVIII or FIX and cyclophosphamide followed by regular treatment with FVIII or FIX, was found to modify antibody titers, suggesting tolerance induction after 2 to 3 weeks.^3,15 

A different approach to inhibitor suppression, known as the “Bonn method,” was introduced by Brackmann and Gormsen.¹⁶  Tolerance was induced by giving patients massive doses of FVIII (200 U/kg b.w. daily) together with an activated prothrombin complex concentrate (APCC) in a dose of 40-60 U/kg to prevent intercurrent bleeding. Following an inhibitor peak within the first months, the antibody titer slowly decreased.¹⁷  A few patients in other centers have successfully been given the same high-dose FVIII treatment.^18,19,20  In the patient reported by White et al¹⁹  no APCC was co-administered. In spite of these successes reported with the long-term high-dose FVIII protocol in hemophilia A patients, the protocol has only been used to a limited extent outside the Bonn center, probably due to the requirement for very large amounts of FVIII concentrate for long periods of time and the consequent high cost. In an effort to control costs, lower dose schedules (25-50 U/kg) have been used with varying success.^21,22 

Results of long-term use of FVIII administration to achieve immunologic tolerance against FVIII in 204 patients treated at 40 hemophilia centers indicated that results seem to be dependent on the use of high-dose protocols (> 100 U/kg/day) and the presence of low levels of inhibitor (< 10 BU/ml) at enrollment.²³  Tolerance was achieved in 107 out of 158 patients (67.7%) who received treatment long enough to judge the outcome, while 39/107 (24.7%) did not respond. Treatment periods of up to 10 months or longer have been reported to be necessary to induce immunologic tolerance.

Unlike hemophilia A patients with inhibitors, patients with hemophilia B present special problems, since the administration of high doses of FIX concentrate is associated with a high risk of anaphylactic reactions.²⁴  Attempts to induce immunologic tolerance have achieved only minimal success and have been associated with the development of nephrotic syndrome.^25,26 

Antibodies against FVIII are relatively species specific, and therefore, porcine FVIII can be used to increase FVIII activity in patients with FVIII inhibitors. Since the early 1980s a highly purified porcine FVIII concentrate has been available and is reportedly effective. Cross-reactivity of antihuman FVIII antibodies to the porcine product has been recognized, and a high antihuman FVIII titer is usually associated with a proportionally increased antiporcine titer. Furthermore, with repeated exposure to the porcine concentrate, patients tend to develop antibodies against the porcine FVIII.²⁷  Large reductions in platelet count were also seen in association with intensive replacement therapy over several days for surgery or trauma.

To find more convenient procedures to treat inhibitor patients, much effort has been focused on finding agents capable of inducing hemostasis independent of the presence of FVIII or FIX. Partially purified plasma-derived concentrates containing all the vitamin K-dependent coagulation factors, prothrombin complex concentrates (PCC) and activated prothrombin complex concentrates (APCC), are still used in spite of a rather low efficacy rate, of about 50-60%.^28,29 

Concurrence as to which factor or factors in the APCCs that are responsible for the hemostatic effect has never been reached. Candidates for the FVIII by-passing effect include activated FVII (FVIIa), FIXa and FXa.^30,31  Also, the material responsible for the thrombo-embolic events being reported in association with the use of PCCs and APCCs^13,32  remained unidentified, although both FXa and FIXa have been discussed as candidates.^13,31,32  Since the APCC and PCC contain trace amounts of FVIII protein, an anamnestic response may be seen in patients given these concentrates.³³ 

FVIIa is the only activated coagulation factor that is not enzymatically active by itself. The requirement for tissue factor (TF) suggests that rFVIIa is hemostatically active only at a local level where TF is available. Using rFVIIa instead of the mixture of activated coagulation factors included in the prothrombin complex concentrates, should theoretically minimize the risk of inducing a disseminated intravascular coagulation and the development of systemic thromboembolic side effect.

TF, a membrane-bound protein present in the subendothelium, is synthesized by cells in the tunica media and in the fibroblast-like adventitial cells surrounding vessels, all anatomically separated from blood.³⁴  Furthermore, it has not been demonstrated to be present in the resting endothelium. Tissue injury disrupts the endothelial cell barrier that normally separates TF from the circulating blood, resulting in the binding of TF to FVII or FVIIa and thereby inducing hemostasis.³⁵  FVII and other vitamin K-dependent coagulation proteins, relatively small proteins, are present in the interstitial fluid,^36,37  where the TF-expressing cells occur.

Tissue injury or inflammation may enhance the hemostatic process by increasing vascular permeability³⁸  or disrupting endothelial cells, and may thereby expose TF at the site of injury. Also, the concentration of FVII at the interstitial TF site may increase. In addition such an injury would increase the availability of platelets at the site of injury. These platelets might become activated by the thrombin formed initially by the TF-FVIIa complex,³⁹  thereby exposing phosphatidylserine and providing the template for full thrombin generation. This process remains compartmentalized and confined to the exposed phosphatidylserine and TF. Although antithrombin does not inhibit free FVIIa, it has been found to inhibit cell surface bound FVIIa-TF complexes,⁴⁰  indicating the important role of antithrombin in the regulation of the initial FVIIa-TF induced hemostasis.

Recently, it was shown that concentrations of FVIIa much higher than those found normally in the circulating blood are able to mediate a TF-independent conversion of FX into FXa on a phospholipid surface.³⁹  The same group showed that rFVIIa binds to activated platelet surface with a low affinity. Thrombin formation, similar to that seen when normal concentrations of all coagulation proteins were present in this cell-based system, was detected in the absence of FVIII or FIX when rFVIIa was added in concentrations of 50 nM or greater.⁴¹  These findings indicate that rFVIIa in concentrations higher than 50 nM may be able to compensate for the absence of FVIII and/or FIX in vitro are consistent with the clinical experience using rFVIIa in hemophilia patients with FVIII or FIX inhibitors.

In a different type of reconstituted coagulation model using TF relipidated in phospholipid vesicles, an inhibitory effect of single-chain FVII on the TF-dependent thrombin generation was demonstrated.⁴²  It was postulated that the hemostatic effect of the administration of exogenous rFVIIa is mediated at least partly by overcoming the inhibition by the zymogen FVII.

Recombinant FVIIa has been used to treat more than 1000 patients and more than 100 000 standard doses have been given. Recombinant FVIIa has been investigated extensively in clinical studies with a 90% efficacy rate.^43,44,45  In a randomized, double blind, multicenter trial there seemed to be a trend towards a better outcome for joint bleeds treated within 6 hours.

To optimize the treatment and to make it more similar to that of hemophilia patients without inhibitors, a home treatment study with rFVIIa was initiated. A fixed dose of 90 μg/kg at 3 hour intervals was begun within 8 hours of the onset of mild-to-moderate bleeding episode. The overall efficacy rate was 92% (452/490 joint bleeds) and the mean number of injections to achieve hemostasis was 2.2.⁴⁶ 

In the past, elective surgery has been contraindicated in hemophilia patients with inhibitors due to the risk of uncontrollable bleeding. However, an open knee joint synovectomy was accomplished under the cover of rFVIIa and tranexamic acid in 1988.⁴⁷  Following this first success, a number of elective major surgical procedures have been performed in hemophilia patients with inhibitors.⁴⁵  A dosing schedule with a bolus dose between 89 and 118 μg/kg given every second hour for 24 hours and thereafter the same dose at increasing intervals for 3-5 days was found to give excellent hemostasis. In patients undergoing minor surgery and dental procedures, the same dosing regimen was used as in major surgery, but the period of treatment was shorter and the intervals between doses were extended earlier with an efficacy rate of 92%.⁴⁴  A prospective, double-blind, randomized trial compared two doses of rFVIIa (35 μg/kg and 90 μg/kg) in hemophilia patients with inhibitors who were undergoing surgery and showed an overall efficacy with satisfactory hemostasis in 23 of the 29 patients. All but one of the five patients who needed extra doses were in the low-dose group. A statistically significant difference in efficacy favoring the high-dose group was observed including number of days of dosing and total consumption of rFVIIa.⁴⁸ 

In most studies, rFVIIa has been administered as single i.v. bolus injections. Some experience, however, also exists with rFVIIa administered by continuous infusion.^49,50,51  However, recently several cases of bleeding have been reported during continuous infusion of rFVIIa.⁵²  In a recent comparison of a single bolus with a continuous infusion, the investigators reported slightly better results with a lower consumption of rFVIIa (300 vs 360 mg/kg) and a higher efficacy (82% vs 72%) after one single bolus dose of 300 μg/kg in joint and muscle bleeds. The authors concluded that a single high dose of rFVIIa seemed to be the most appropriate treatment of hemarthrosis and muscle bleeds.⁵¹ 

Since rFVIIa enhances the thrombin generation on surfaces with exposed TF or on activated platelet surface with exposed phosphatidylserine, it can also induce hemostasis in other hemostatic disorders. Such a hemostatic effect has been observed in patients with platelet defects, such as Glanzmann's and Bernard-Soulier's thrombasthenia.⁴⁵  Recently, a hemostatic effect of single doses of rFVIIa was observed in a patient with a life-threatening profuse traumatic bleeding.⁵³ 

According to the clinical experience of rFVIIa in acute bleeding and in surgery, a dose of 90 μg/kg given as a bolus every 2^nd hour at least for the first 24 hours is effective, with increasing intervals of 3-6 hours thereafter. This dose may be adjusted according to clinical response and the type of surgery. The same dose was effective in the treatment of minor-to-moderate joint and muscle bleeding episodes in a home setting requiring a mean number of injections of 2.2 per bleed.

Thrombin is essential for hemostasis because it activates FVIII, FV, FXI and FXIII. The rate of thrombin formation is of importance for the fibrin structure, which is essential for the stability of the hemostatic plug.^54,55 

As pointed out by Monroe et al,³⁹  full thrombin generation on the activated platelet surface in the absence of FIX could be induced by adding extra rFVIIa. At FVIIa concentrations of 50 nM or higher, thrombin generation approached that seen in the presence of FIX/FVIII, and normalization of the thrombin generation was achieved after the addition of 150 nM of rFVIIa.⁴¹ 

A dose of 90-100 μg/kg of rFVIIa (specific activity 50 U/μg) roughly corresponds to plasma levels of about 2 μg/ml of rFVIIa (60-90 U/ml; 50 nM), indicating that this dose may be enough to induce thrombin generation within the range found in normal individuals.⁵⁶  However, is should be kept in mind that the clearance rate of rFVIIa does vary and that, in children below 15 years of age, it may be three times the value found in adults.⁵⁷  Since it is not known for how long the FVIIa level needs to be above 50 nM to induce the formation of a stable and solid hemostatic plug, shorter clearance in some individuals may contribute to a less optimal hemostatic effect of rFVIIa. This finding may justify the use of much higher doses.

On the contrary, van't Meer et al⁴²  observed that 10 nM of FVIIa normalized the thrombin generation profile in their in vitro model, which would correspond to a FVII:C plasma level of 6-10 U/ml. However, several of the patients receiving continuous infusion of rFVIIa with a plasma level of FVII:C of 10 U/ml did require extra bolus doses due to rebleedings.⁴⁹  No such bleedings occurred when bolus dosing (90 μg/kg b.w.) was used (FVII:C levels of at least 30 U/ml).⁴⁸  Thus, peak plasma levels of 30 U/ml or higher of FVII:C seem to be required to achieve a reliable hemostasis. Furthermore, the use of higher doses of rFVIIa in the range of 200 to 320 μg/kg (corresponding to 100 or 150 nM of FVIIa) given as a single bolus has resulted in excellent hemostasis,^51,57  which indicates that higher initial thrombin peaks are important for the formation of a stabile fibrin hemostatic plug. Orthopedic traction of bilateral knee joint contractures was performed in one hemophilia B patient under the cover of rFVIIa in a dose of 240 μg/kg given 4 times (trough level of FVII:C was around 25 U/ml) during the first 48 hours and 3 times (trough level of FVII:C was around 13 U/ml) during the following week of active traction. Thereafter, rFVIIa was administered in the same dose twice per day for a month during the casting and wedging. During the following month with extensive physiotherapy he got one dose of 240 μg/kg per day. No bleeding episodes occurred in the knee joints during the entire treatment period indicating a preventive effect of rFVIIa in a situation usually associated with complicating joint bleedings.⁶² 

At plasma levels of FVII:C of above 30 U/ml the one-stage clotting assay (FVII:C) satisfactorily reflects the level of FVIIa, provided the patient plasma is diluted so that the plasma level is between 0.5 and 1 U/ml, which corresponds to the levels found in normal plasma without any administration of rFVIIa. An excellent correlation was demonstrated between the levels measured with the standardized FVII:C one-stage clotting assay⁵⁸  and those obtained in a modified assay of FVIIa principally according to Morrissey et al.⁵⁹ 

In studies including elective surgery patients^45,48  and treatment of joint and muscle bleedings in a home setting⁴⁶  no serious side effects were recorded. One patient in the home treatment study experienced a superficial thrombophlebitis at an intravenous access site.

Despite the fact that a number of patients with life-and limb-threatening bleeding episodes had proven septicemia and a number of the patients had known cardiovascular disease, only one patient has been reported to develop consumption coagulopathy when on rFVIIa treatment.⁶⁰  In this patient, the consumption reversed, despite continued rFVIIa treatment. Several other patients with significant infections or septicemia received treatment with rFVIIa with no signs of consumption coagulopathy.

Out of the more than 100,000 standard doses of rFVIIa given, two patients (57 and 81 years old) with non-fatal deep venous thrombosis (DVT) have been reported. Three patients developed transient cerebral ischemic events. In total, six patients (three hemophilia A and 3 acquired hemophilia) have had acute myocardial infarctions (5 above 67 years old). Two patients had diabetes type II and two others a previous history of cardiovascular disease.

No antibodies against FVII have been found.⁶¹ 

Katherine A. High, M.D.^*

Clinical gene therapy began ten years ago with the treatment of two young girls with SCID due to ADA deficiency. In the ensuing decade, over 4,000 patients have been treated on gene transfer protocols, but success has been slow in coming. Within the last year, however, Fischer and colleagues have reported, first at last year's ASH meeting and subsequently in Science,¹  the successful treatment of two boys with X-linked SCID by a gene transfer approach in which hematopoietic stem cells were transduced with a retroviral vector expressing the common γ-chain of several cytokine receptors. In addition, two other groups have reported evidence of a therapeutic response in cancer patients treated in phase III trials with plasmid-based or adenoviral vectors.^2,3  Thus we can anticipate the entrance of gene transfer products into the therapeutic armamentarium within the next several years.

Interest in developing a gene transfer approach to the treatment of hemophilia continues to be high, and three clinical trials are already underway. Experience with prophylactic regimens of protein concentrates over the last 30 years has established that continuous maintenance of circulating levels of clotting factor > 1 % is adequate to prevent most of the mortality and much of the morbidity associated with the disease;⁴  these data provide a strong rationale for the potential for success of a gene-based approach. Compared to other genetic diseases, hemophilia has a number of characteristics that are likely to facilitate the development of a gene transfer approach to treatment. Biologically active clotting factors can be synthesized in many different cell types, so that there is latitude in choice of target cells. The therapeutic window is wide, since factor levels as low as 1.5% of normal are likely to improve the clinical symptoms of the disease, and levels of 100% are still within normal limits. There are large and small animal models of the disease (genetically engineered mice and naturally occurring dog models), and the murine and canine FVIII and FIX genes have been cloned and are available, allowing detailed feasibility studies before moving to clinical trials. Finally determination of therapeutic efficacy is straightforward in the case of hemophilia, since circulating levels of clotting factor are easy to measure and correlate well with clinical manifestations of the disease.

A majority of the clinical trials of human gene therapy have made use of retroviral vectors, and early on there was considerable interest in using retroviral vectors introduced into liver to treat hemophilia. In 1993, Kay et al demonstrated long-term expression of canine FIX in dogs that had undergone two-thirds partial hepatectomy followed by splenic vein infusion of a retroviral vector expressing canine FIX under the control of a liver-specific promoter.⁵  Despite the antecedent partial hepatectomy, transduction efficiency with the retrovirus was low, as were levels of expression (4 ng/ml, < 0.1%), but the study demonstrated unequivocally that long-term expression could be achieved. Partial hepatectomy, carried out to induce cell division in the remaining hepatocytes (a prerequisite for retroviral transduction), is unappealing as an approach to human therapy, and subsequent work has focussed on other methods of inducing cell division in hepatocytes. One approach for inducing hepatocyte replication in the setting of liver-directed gene therapy has been portal branch occlusion.^6,7  Branches of the portal vein are ligated, resulting in apoptosis of hepatocytes in the occluded lobes and compensatory replication of hepatocytes in the nonoccluded lobes. In rats and pigs, this approach has resulted in levels of reporter gene expression that are 20-50% of levels attained with partial hepatectomy and is associated with considerably lower morbidity and mortality in the rat model. In an alternate and less invasive approach, several groups^8,9  have demonstrated that antecedent infusion of recombinant keratinocyte growth factor (KGF) or hepatocyte growth factor (HGF) can increase the number of replicating hepatocytes into the range of 5-13% and increase the number of transduced hepatocytes to 1-2% (from an untreated baseline of ∼0.1% or less). The use of high titer retroviral preparations results in even higher levels of transduction, and higher levels can also be achieved through direct portal vein infusion of high doses of growth factors. Finally, another approach is simply to take advantage of the higher rate of hepatocyte replication in newborn and juvenile animals. Thus VandenDriessche et al have recently demonstrated therapeutic levels of FVIII in newborn mice with hemophilia A.¹⁰  These investigators prepared high titer vector (10⁹-10¹⁰ cfu/ml) expressing human FVIII and infused dose of 4 × 10⁹ cfu/kg intravenously; 6/13 mice expressed FVIII, with 4 of these expressing at levels > 50% normal human plasma levels (measured as FVIII activity, not antigen). The remaining mice developed antibodies to human FVIII, but this adverse event might have occurred in fewer animals if a species-specific transgene had been used. These data are consistent with results reported earlier in preliminary form by Greengard and colleagues in which infusion of high titer retrovirus into juvenile animals resulted in therapeutic levels of FVIII in rabbits and dogs,¹¹  although it should be noted that estimation of levels in this case was complicated by the presence of antibodies to the human FVIII transgene product. Key factors in the promising results of VandenDriessche et al¹⁰  appear to be the rate of hepatocyte proliferation in newborn mice and the use of very high titer vector. Whether these encouraging results in young animals can be extended to adults is currently being evaluated in a phase I clinical trial (Chiron Corporation) where highly concentrated retroviral vector, pseudotyped^* with an amphotropic envelope and expressing B domain-deleted FVIII, is infused intravenously into adults with severe hemophilia A¹²  at doses ranging from 1 × 10⁸ cfu/kg to 8 × 10⁸ cfu/kg. Experimental evidence from a number of groups indicates that some baseline level of hepatocyte transduction occurs in adult animals even in the absence of a stimulus to hepatocyte replication,⁹  but it is not entirely clear which cells are transduced in this procedure. As of June 2000, ten subjects with severe hemophilia A had been enrolled in this trial, with expected total enrolment of ∼20 subjects. No serious adverse events have been reported among those enrolled, and the trial is continuing. Determining whether the level of transduction achieved will be adequate to result in therapeutic levels of transgene expression in humans will be a goal of this trial.

Lentiviral vectors,¹³  a newer gene delivery vehicle based on the human immunodeficiency virus (HIV), have also been shown to transduce liver, muscle and hematopoietic cells, and thus could potentially be used for gene therapy for hemophilia. Work published three years ago by Kafri et al¹⁴  demonstrated stable expression (22 weeks) of a humanized green fluorescent protein (GFP) following direct intraparenchymal injection of a lentiviral vector. The authors showed that the percentage of cells transduced at the site of injection was high (∼90%), comprising up to 3-4% of hepatocytes, and that vector could be successfully re-injected. Park et al have recently reported that antecedent partial hepatectomy increases lentiviral transduction efficiency by ∼30-fold in mouse hepatocytes following infusion of vector into the portal vein; in addition, in co-labeling experiments with bromodeoxyuridine (BrdU) to detect DNA synthesis and a lentiviral vector expressing lacZ, it was shown that ∼90% of cells expressing lacZ were co-labeled with BrdU. These findings raise the possibility that DNA synthesis may be required for efficient lentiviral transduction in vivo in mice.¹⁵  The demonstration that dose-dependent increases in serum transaminases occurred with lentiviral vector infusion suggests a mechanism (liver injury) for increased cell cycling.¹⁶  Park et al have carried out studies using a lentiviral vector expressing FIX under the control of an EF1α promoter to direct sustained expression of the transgene following introduction of vector into the portal veins of C57B1/6 mice.¹⁶  Consistent with their earlier studies, preparative partial hepatectomy resulted in a 4- to 6-fold increase in levels of expression compared to non-hepatectomized mice. Similar experiments attempted with a FVIII transgene resulted initially in levels of expression of ∼36 ng/ml (18% of normal plasma levels) but FVIII levels became undetectable by week 8 because of the development of anti-FVIII antibodies. There are as yet no published data using lentiviral vectors in the canine model of hemophilia, although Naldini and colleagues have reported such experiments in preliminary form.¹⁷  Again, levels of expression could not be determined as the treated dog developed inhibitory antibodies to the canine transgene product (see below).

Lentiviral vectors have not yet been used in human trials. Work by Naldini and colleagues¹⁸  has been directed at minimizing the possibility of generating replication-competent recombinants (RCR) through the use of a conditional packaging system that uses only a fractional set of HIV genes. The recent development of a sensitive assay for RCR (which can detect as little as 1 fg of p24) should also be useful for eventual clinical testing of these vectors.

Adeno-associated viral (AAV) vectors are engineered from a parvovirus that is not associated with any known human disease and are capable of directing sustained expression of a transgene following introduction into muscle, liver or CNS. The pre-clinical data demonstrating the efficacy of AAV-mediated gene transfer for hemophilia B are arguably the most solid in the field; they constitute the only published data demonstrating long-term correction of clotting parameters such as the aPTT in a large animal model of hemophilia. The safety and efficacy of this approach in humans are currently being tested in a trial in which vector expressing human FIX is introduced into intramuscular sites. Additional trials in which vector will be introduced into hepatocytes via an intravascular approach are currently in the planning stages; there are at this time no approved clinical trials using AAV in liver, and initiation of such trials awaits completion of the required safety studies.

The original reports of long-term expression of both secreted and non-secreted gene products following introduction of AAV vector into muscle appeared in 1996.^19,20  In the following year, sustained expression of therapeutic levels (250-350 ng/ml, 5-7% of normal human plasma levels) of FIX was demonstrated following introduction of an AAV vector expressing FIX into intramuscular sites in the hind limbs of mice.²¹  An important test of the feasibility of a gene therapy strategy for hemophilia is scale-up to the canine model; strategies that do not demonstrate efficacy in the canine model are unlikely to be successful in humans. In the case of AAV-mediated, muscle-directed gene transfer for hemophilia, extension to the canine model allowed investigators to address a number of questions that could not be assessed in studies in inbred strains of mice, including feasibility of scale-up of vector production, efficiency of transit of FIX from the site of synthesis in the muscle to the circulation, and characterization of the immune response to the transgene product. Herzog et al introduced an AAV vector expressing canine FIX into the hind limbs of dogs weighing between 6 and 20 kg, and demonstrated long-term, dose-dependent expression of biologically active FIX.²²  All dogs showed partial correction of the whole blood clotting time and dogs treated at doses at or above 3 × 10¹² vector genomes (vg)/kg manifested partial correction of the aPTT as well. Expression of clotting factor has now lasted for over three years since the time of the initial (and only) injection, with the experiment still ongoing. One dog showed development of a transient inhibitory antibody; this disappeared without any specific therapy and has not reappeared on rechallenge with normal canine plasma (containing canine FIX). Ongoing experiments in hemophilic dogs have been aimed at defining parameters that enhance or reduce the risk of inhibitor formation. These data suggest that the dose of vector per site is a better predictor of inhibitor formation than the total dose or the dose/kg, although the latter has some effect on risk.²³  These and other pre-clinical studies form the basis for the current clinical trial in patients with severe hemophilia B being conducted at The Children's Hospital of Philadelphia and at Stanford University. The trial involves ultrasound-guided intramuscular injection of an AAV vector expressing human FIX, under the control of a CMV promoter/enhancer. The phase I/II trial is designed to include 9-12 patients and will assess the safety of the approach, in particular, the possibility of inhibitor formation, the levels and duration of expression in three dose cohorts, and the risk of germline transmission of vector sequences. Safety data on subjects in the low- and mid-dose cohorts (2 × 10¹¹ vg/kg and 6 × 10¹¹ vg/kg, respectively) have shown no evidence of inhibitory antibody formation or of germline transmission of vector, and muscle biopsies have shown evidence of gene transfer and expression in those treated. Initial data from this trial, the first gene transfer protocol in which rAAV has been introduced parenterally, were published earlier this year.²⁴ 

Experimental evidence in support of an AAV-mediated liver-directed approach to treatment of hemophilia continues to grow and now constitutes the strongest data in the literature in terms of levels of FIX achieved in large animal models. In mice, Snyder et al had demonstrated sustained expression of FIX at levels > 1000 ng/ml following infusion of an AAV vector expressing human FIX (hFIX) under the control of an MFG promoter into the portal vein.²⁵  Similar results were subsequently reported by Nakai et al, using an AAV vector with a eukaryotic housekeeping promoter (EF1α) and by Wang et al using a vector with a liver-specific promoter^26,27  introduced into the portal veins of mice. Snyder et al subsequently extended these findings to the hemophilic dog model where they showed long-term partial correction of the whole blood clotting time and the aPTT following introduction of an AAV vector expressing canine FIX into the portal veins of dogs with severe hemophilia B.²⁸  Similar results have been described by Verma and colleagues,²⁹  and Herzog et al have now achieved levels of 600-700 ng/ml (12-14% normal plasma levels) in dogs with hemophilia B.³⁰  Based on these results, and on the safety results from parenteral administration of rAAV in skeletal muscle, several groups are now developing protocols for AAV-mediated, liver-directed human gene therapy trials for hemophilia B.

Efforts are also underway to extend the use of an AAV-mediated, liver-directed approach to hemophilia A, where the size of the transgene (the 4.4 kb B domain-deleted FVIII construct) presents an obstacle. Because of the constraints on the size of the insert that can be accommodated by the AAV vector (∼5 kb), several novel strategies have been devised to allow expression of FVIII from an AAV vector. In the first of these, described by Couto et al, two vectors are constructed, one expressing the heavy chain (A1 and A2 domains) and the other the light chain (A3, C1 and C2) of FVIII. Following introduction of vectors into the portal circulation, biologically active FVIII is produced in the circulation at supraphysiologic levels (200-400 ng/ml), presumably from hepatocytes that are co-transduced with both vectors.³¹  Chao et al have approached the problem differently by constructing a single vector with a small promoter. Using a minigene consisting of BDD FVIII driven by the thymidine kinase promoter linked to a hepatitis B enhancer, this group showed FVIII levels of 55 ng/ml in NOD/SCID (non-obese diabetic/severe combined immunodeficiency) mice following portal vein injection of vector at a dose of 6 × 10¹² viral particles/kg.³²  A third strategy, proposed by Engelhardt and colleagues, takes advantage of the molecular configuration of recombinant AAV within a transduced cell. Since the vector genome is present within transduced cells as head-to-tail concatemers (restriction digests indicate that the vector DNA is present in concatameric form within the cell, with individual gene cassettes arranged end to end in a head-to-tail fashion, i.e. the 3′ end of one unit is joined to the 5′ end of the next and so forth), two vectors can be constructed, one containing regulatory elements and a splice donor, the other containing a splice acceptor, the transgene of interest, and a polyadenylation signal.³³  Whether this strategy will be successful for AAV-mediated expression of FVIII remains to be seen.

Both the liver-directed and muscle-directed approaches have shown efficacy in the canine model, and it is not yet clear which will be more useful clinically. There is a clear dose advantage in favor of liver, on the order of one to two logs with currently available constructs; this is likely accounted for primarily by the more efficient transit of FIX into the circulation from the hepatocyte compared to transit from skeletal muscle fibers. In addition, it is likely that all post-translational modifications will be executed accurately and efficiently in hepatocytes, whereas some modifications affecting FIX recovery are not efficiently performed in skeletal muscle.³⁴  On the other hand, introduction of vector into skeletal muscle can be done by simple IM injections, while introduction of vector into the liver will require an invasive procedure in which a catheter is introduced into the hepatic artery. An additional uncertainty regarding the liver-directed approach is the effect of underlying hepatitis on vector transduction, and vice versa (see below). Among adults with severe hemophilia, over 90% have been exposed to hepatitis B and over 80% to hepatitis C.^{35,36,37,38,39}  The effect of ongoing inflammation and altered cytokine profiles within the target tissue may increase the likelihood of inhibitor formation in the setting of gene transfer. Finally, if sites of injection into skeletal muscle are judiciously selected, one could conceivably reverse the procedure by resecting the injected sites if some unanticipated adverse event were to occur; introduction of vector into the liver is, on the other hand, an irreversible event. Given these considerations, it is appropriate that skeletal muscle was selected as the initial site for parenteral injection of AAV vector, but liver may eventually be the best target, especially for individuals who are not infected with hepatitis. At this point, the best course of action is to continue efforts to develop clinical approaches using both of these target tissues. It may be that both will be useful but will have different indications, e.g. depending on presence of hepatitis or other factors.

Adenoviral vectors have a number of attractive features as gene delivery vehicles, including ease of preparation and efficient transduction of liver following introduction of vector into the peripheral circulation. These characteristics were exploited by Kay and coworkers to obtain high-level expression of canine FIX in hemophilic dogs as an early proof of principle for this approach.⁴⁰  However, expression was short-lived, and work by many other groups has now established that the immune response to the vector, characterized by a cytotoxic T lymphocyte response against cells harboring the vector, will make it problematic to obtain long-term expression using early generation adenoviral vectors.^41,42 

Connelly et al have explored the use of earlier generation adenoviral vectors as an approach to treating hemophilia A.^{43,44,45,46,47}  Using an adenoviral vector expressing B domain-deleted FVIII, these workers were able to demonstrate phenotypic correction of the bleeding diathesis in mice with hemophilia A.⁴⁷  Levels of expression were initially > 2000 mU/ml and, as expected, declined gradually over 9 months to ∼ 100 mU/ml; lower doses of vector resulted in longer-term expression, presumably due to a less vigorous immune response. Attempts to extend this approach to the canine model have been hampered by hepatotoxicity and inhibitor formation, the latter perhaps linked to the former (see below).⁴⁸  Work by a number of groups has been directed to the task of deleting all of the adenoviral backbone genes with the goal of diminishing toxicity and prolonging expression.^{49,50,51,52,53}  Zhang and colleagues⁵⁴  have described the use of a fully deleted adenoviral vector to correct the phenotype of mice with hemophilia A. Using a construct containing the full-length human FVIII cDNA under the control of a 12.5 kb albumin promoter, this group has demonstrated long-term expression of therapeutic levels of FVIII (100-800 ng/ml) in 3/16 hemophilic mice. The remaining 13 mice developed antibodies to human FVIII within 3-8 weeks of vector administration, so that FVIII levels could no longer be measured. Administration of the vector was not associated with any evidence of hepatotoxicity in the mice, but the same was true of earlier generation adenoviral vectors, where hepatotoxicity in mice was much less marked than in dogs.⁴⁸  One of the uncertainties regarding hepatotoxicity of adenoviral vectors is whether the immune response is triggered by adenoviral proteins (e.g. hexon and fiber proteins) alone, or whether low-level ongoing viral replication is required for the harmful immune response. A clear understanding of the underlying cause of the immune response will have implications for the safety of a gutted adenoviral vector, particularly in the setting of hemophilia where many potential subjects are infected with hepatitis. In addition to the effect on duration of expression, a brisk immune response to the vector may facilitate the development of a harmful immune response to the transgene product, as has been demonstrated to occur for earlier generation adenoviral vectors.⁵⁵  This type of response is best evaluated in a hemophilic animal model, which is less tolerant to the transgene product (and thus a more faithful model of humans with the disease) than a normal animal. Two other issues that will need to be addressed before gutted adenoviral vectors are used clinically are the degree of contamination of clinical grade preparations by helper virus and the existence of a threshold effect, which has been described by Connelly and coworkers for earlier generation adenoviral vectors, in which the dose-response is linear down to a certain dose, below which there is no expression.⁵⁶  The existence of this phenomenon would have implications for dosing in a clinical trial.

A novel approach to gene therapy is the use of chimeric RNA/DNA oligonucleotides to activate endogenous repair systems resulting in in situ correction of point mutations.^57,58  Based on the observation that RNA/DNA hybrids are active in homologous pairing reactions in vitro, this strategy makes use of a five-nucleotide DNA stretch containing the wild-type sequence, flanked on either side by ten modified RNA residues and a 3′ G:C clamp. The oligonucleotides are infused intravenously (complexed to a protecting polycation, polyethyleneimine, which in turn is modified with a ligand, lactose, that serves to target the oligonucleotides to hepatocytes via the asialoglycoprotein receptor). Once in the nucleus, the unpaired nucleotide is recognized by endogenous repair systems resulting in alteration of the DNA sequence of the targeted gene. The efficiency of this event is both dose dependent and sequence dependent; thus for example efficiency for the sequence encoding the mutation in the bilirubin UDP-glucuronosyl-transferase gene in the Gunn rat, a model of Crigler-Najjar disease in humans, is quite high.⁵⁹  Convincing evidence of correction of a hemophilic point mutation has not been obtained. In general however, it should be noted that this strategy may not be well-suited for application to hemophilia, since the disease mutations are quite heterogeneous⁶⁰  (each mutation would require a different oligonucleotide), and many patients, especially with hemophilia A, have disease due to large gene inversions that could not be readily addressed by this method. In addition, since the efficiency of gene conversion appears to be sequence specific, the heterogeneity of the patient population at the nucleotide level could lead to variable results.

In yet another strategy, Transkaryotic Therapies, along with investigators at the Beth Israel Deaconess Medical Center in Boston, are carrying out a phase I study in which autologous fibroblasts transfected with a B-domain deleted FVIII plasmid are re-implanted on the omentum in subjects with severe hemophilia A. Pre-clinical studies in support of this work include studies in immunodeficient mice in which clonal rabbit skin fibroblasts transfected with a BDD hFVIII were implanted on the omentum of the mice.⁶¹  BDD hFVIII was detected in the plasma of these animals by ELISA and by Western blot, although levels are not yet reported. There are no reported experiments using this approach in hemophilic animals, so that the immune response to the transgene product in this setting is unknown. For the clinical trial, the BDD hFVIII is under the control of a eukaryotic promoter that allows optimal expression in fibroblasts. In the phase I study, two dose cohorts will be tested, consisting of 10⁸ or 4 × 10⁸ transfected autologous fibroblasts. The trial began in December 1998 and is currently ongoing, with six subjects enrolled as of June 2000. No serious adverse events have been reported among those enrolled in the trial, and a decrease in factor usage has also been reported for some subjects.⁶²  This ex vivo approach is intriguing and the expected publication of the pre-clinical data supporting the trial should make it easier to assess the likelihood of long-term expression at adequate levels.

Plasmids have a number of advantages as gene transfer vehicles, including ease of production and absence of some of the safety concerns associated with viral vectors, e.g. harmful immune responses to the vector itself and potential for generation of replication-competent forms. A major limitation of plasmid-mediated gene transfer, however, has been the inability to achieve long-term expression because of failure of the transferred DNA to integrate into the host cell. Yant and coworkers have recently described the use of a synthetic transposable element derived from a fish transposon (transposons are naturally occurring genetic elements capable of moving from one chromosomal location to another) to enhance integration of plasmid-encoded FIX into hepatocytes of normal and hemophilic mice.⁶³  Tail vein injection of a plasmid encoding the synthetic transposase along with a plasmid encoding FIX resulted in long-term expression at levels in the range of 200-400 ng/ml (4-8% normal circulating levels). Moreover expression could be enhanced as much as 4-fold by re-injection of both plasmids. This system thus seems to afford the possibility of stable non-viral gene transfer. It will be of interest to determine how efficiently it can be scaled up to larger animal models.

Leong et al⁶⁴  have presented a preliminary report in which FIX plasmid DNA contained within a chitosan-DNA (a commercially available chitin-derived biopolymer, chemical name N-acetyl-D-glucosamine, which has been used in a number of drug delivery applications, including delivery of DNA) nanosphere is embedded within gelatin cubes and fed to mice at a dose of 25 μg plasmid in a single treatment. Treated mice showed levels of 45 ng/ml (∼1 % normal plasma levels), although levels gradually declined to undetectable levels over a 14-day period. The effects of repeated oral administration are under study. Although many questions remain (e.g., site of synthesis, biological activity of the protein, ability to achieve adequate levels in a larger animal), this study raises the exciting possibility of a non-invasive method for achieving gene transfer.

Determination of clinical efficacy has been a source of difficulty for some diseases treated through a gene therapy approach, but is unlikely to present a problem in the case of hemophilia. A wealth of clinical and laboratory data document that plasma FVIII and FIX activity levels correlate well with clinical severity of disease. As noted above, the Swedish prophylaxis studies,⁴  using as a goal maintenance of trough factor levels of > 1%, have documented prevention of joint disease and of life-threatening bleeds at this level. Thus, levels of 1-5% are likely to result in considerable improvement of disease phenotype, and levels of 5-20% are likely to free patients from any substantial need for clotting factor concentrates except in the setting of surgery or trauma. It is unclear at this time whether it is prudent to aim for normal levels without incorporating a control element^{65,66,67,68,69}  that would allow for downregulation of expression in the event of an overshoot. The use of secondary endpoints such as total factor usage and number of bleeding episodes will also be of interest but are unlikely to be considered convincing in the absence of measurable activity levels in the circulation. An exception to this would be strategies that are targeted to specific cell types (platelets, synovial cells) that are abundant at bleeding sites but unlikely to result in systemic levels of expression.

An important issue facing all gene therapy trials for hemophilia is the risk of forming inhibitory antibodies to the transgene product. Currently, formation of neutralizing antibodies, or inhibitors, is the most common complication of protein-based therapy, occurring in ∼20% of patients with hemophilia A and ∼3% of those with hemophilia B.^70,71  Despite years of study, it is still not possible to predict with certainty which patients will develop inhibitory antibodies in the setting of protein-based therapy, but certain risk factors have been identified, including the nature of the underlying mutation in the clotting factor gene, inherited characteristics of the individual's immune response, and the circumstances surrounding exposure to the clotting factor protein (i.e. presence of tissue injury or inflammation).

It is likely that there are differences in antigen presentation of clotting factor epitopes in gene-based vs. protein-based treatment. In protein-based therapy, antigen presentation occurs primarily in the setting of MHC Class II determinants, which display peptides derived from proteins taken up from the environment. In the setting of gene therapy though, antigen presentation also occurs through MHC Class I, which presents peptides derived from proteins synthesized within the cell that displays them. Thus gene therapy, which results for the first time in endogenous synthesis of the wild-type protein in a recipient, may be characterized by a different immune response to the transgene product compared to responses seen in protein-based treatment.

Factors that are likely to influence inhibitor formation in the setting of gene therapy include the vector itself, the target tissue used, dose of vector, and inclusion of tissue-specific promoter elements. Vectors that elicit a strong immune response to the viral proteins (e.g. adenoviral vectors) are more likely to elicit an immune response to the transgene product as well.⁷²  Target tissues rich in professional antigen presenting cells might also predispose to inhibitor formation. In the case of adenoviral vectors, evidence suggests that inclusion of tissue-specific regulatory elements that restrict expression to the target tissue alone (and thus do not allow expression in antigen-presenting cells) may reduce formation of inhibitory antibodies.⁷³  Whether this will be generally true is not yet clear. Finally, it is possible that transient immunomodulation at the time of vector administration may block formation of inhibitory antibodies.⁷²  Although a clear understanding of the immunologic mechanisms underlying antigen presentation and immune response in the setting of gene therapy is not yet available (the same may be said of the current protein-based method of treatment), a few comments can be made. First, it is clear that each therapeutic system, consisting of a specific transgene, vector, and target tissue, constitutes a different problem. It will in general not be accurate to make statements about any one of these without defining the other parameters as well. Second, the immune response will be influenced by characteristics of the individual patient, including the underlying mutation, which will determine the level of tolerance to the transgene product, and inherited characteristics of the immune response, which will determine details of antigen processing and presentation in the recipient. These facts underscore the critical importance in the case of genetic diseases like hemophilia of carrying out preliminary experiments in animal models of the disease, where tolerance to the transgene product may be lacking or altered, as is the case for humans with the disease. A more comprehensive discussion of immune response to the transgene product in the setting of gene therapy for hemophilia is available.⁷⁴ 

Among adults with severe hemophilia, the prevalence of HIV infection is high, ∼82% for patients with hemophilia A and 48% for those with hemophilia B,⁷⁵  as is the prevalence of exposure to hepatitis B (90% HbsAbg+)³⁵  and hepatitis C (80%).^36,37,38,39  These prevalent co-morbid conditions are likely to have implications for gene therapy for hemophilia. For example, the currently recommended anti-retroviral therapy (HAART) is likely to block transduction with retroviral and lentiviral vectors, but stopping the medication is generally contraindicated since it may facilitate emergence of resistant strains (although this is a rapidly evolving area, and brief interruption of therapy in individuals who are fully suppressed [< 50 copies virus/ml] may be acceptable). Similarly, the effect of co-existing hepatitis on liver-directed gene therapy is unknown. It is possible that the presence of ongoing inflammation in the liver may predispose to inhibitor formation; alternatively, patients with hepatitis C may be less likely to mount an immune response. Efforts to assess this question may be facilitated by the recent development of animal models of hepatitis infection.⁷⁶ 

For many of the viral and non-viral treatment strategies under consideration, successful gene transfer will result in integration of the donated transgene at random sites within the recipient genome. The long-term consequences of such events are unknown, but the concern exists that integration into a crucial gene sequence, e.g. a tumor suppressor gene, could result in inactivation of the critical gene and an increased likelihood of malignant transformation. Data specifically addressing this point continue to accrue as the number of long-term survivors of therapy with integrating vectors increases. It should be recognized, however, that extended periods of follow-up will be required to determine whether any substantial risk attaches to the theoretical concerns regarding insertional mutagenesis. A conservative approach, until more data are available, would be to limit initial trials with integrating vectors to older subjects who have relatively fewer years of potential life remaining. Such an approach would allow accrual of additional information on effects of integrating vectors while protecting the population who would be most affected by late-appearing complications.

A potential consequence of gene therapy is the introduction of foreign DNA into the gonads of recipients and, potentially, into the germ cells of these individuals.^77,78  This could result in the transmission of the donated gene sequences to subsequent generations. If this resulted in permanent correction of the genetic defect, it could hardly be viewed as a deleterious side effect, but since most vectors integrate randomly, the concern exists that the donated sequences may result in harm to the offspring if, for example, the site of integration disrupts a critical gene sequence in the developing embryo,⁷⁹  or if expression of the donated gene sequence somehow disrupts the normal program of development. Thus, a part of the safety assessment of every gene therapy strategy is a determination of the likelihood that donated gene sequences will be transmitted to future generations. In general these risks are lower for ex vivo strategies than for in vivo gene transfer; there has been no evidence of germline transmission of vector sequences in the ongoing hemophilia trials, two of which use in vivo gene delivery. It will be important to continue to collect data in this area.