After successful efforts in adeno-associated virus (AAV) gene addition for hemophilia B gene therapy, the development of valoctocogene roxaparvovec (Roctavian; Biomarin) over the past decade represents a potential new hemophilia A (HA) treatment paradigm. Roctavian is the first licensed HA gene therapy that was conditionally approved in Europe in August 2022 and approved in the United States in June 2023. Beyond Roctavian, there are ongoing pivotal trials of additional AAV vectors for HA, others that are progressing through preclinical development or early-phase clinical trial, as well as non-AAV approaches in clinical development. This review focuses on the clinical development of Roctavian for which the collective clinical trials represent the largest body of work thus far available for any licensed AAV product. From this pioneering clinical development, several outstanding questions have emerged for which the answers will undoubtedly be important to the clinical adaptation of Roctavian and future efforts in HA gene therapy. Most notably, unexplained year-over-year declines in factor VIII (FVIII) expression after Roctavian treatment contrast with stable FVIII expression observed in other AAV HA gene therapy clinical trials with more modest initial FVIII expression. This observation has been qualitatively replicated in animal models that may permit mechanistic study. The development and approval of Roctavian is a landmark in HA therapeutics, although next-generation approaches are needed before HA gene therapy fulfills its promise of stable FVIII expression that normalizes hemostasis.
After successful efforts in adeno-associated virus (AAV) gene addition for hemophilia B gene therapy,1,2 the development of valoctocogene roxaparvovec (Roctavian; Biomarin)3-7 over the past decade marks a new therapeutic drug class for hemophilia A (HA) treatment. Roctavian is the first licensed HA gene therapy and was conditionally approved in Europe in August 2022 and the United States in June of 2023. In addition to Roctavian, there are ongoing pivotal trials of additional AAV vectors for HA that include, giroctocogene fitelparvovec (Pfizer), dirloctocogene samoparvovec (Spark Therapeutics), and others progressing through early-phase clinical trial or preclinical development (supplemental Table 1). Although most HA gene therapy efforts use AAV-mediated gene addition, other approaches are also in development.8-11 AAV-based therapies for HA target episomal, exogenous expression of B-domain deleted factor VIII (FVIII) in hepatocytes.6,12-15 However, bioengineered FVIII variants that improve expression13,16,17 or function,18,19 analogous to the successful universal adoption of FIX Padua for hemophilia B gene therapy efforts,12,20,21 are being pursued in clinical and preclinical investigation.
HA has long been an attractive monogenic disorder for gene-based therapy, in large part because of the strong correlation between FVIII activity and phenotype, making FVIII activity an excellent biomarker. Most bleeding sequalae and spontaneous hemorrhage occurs in patients with severe (FVIII activity of <1% of normal) or moderate (FVIII activity of 1% to <5% of normal)22 HA. The mild HA phenotype (FVIII activity of 5% to <40% of normal) is heterogenous wheras natural history studies identified FVIII activity of 12% to 20% of normal protect against spontaneous hemorrhage23 and joint bleeds.24 Current HA standard-of-care include recurrent intravenous exogenous FVIII infusion 2 to 4 times per week or subcutaneous administration of the FVIII mimetic, emicizumab (Hemlibra; Genentech), 1 to 4 times per month25-27 to prevent bleeding. Practice guidelines recommend targeting FVIII troughs of ≥5% of normal,28 and the relative FVIII equivalency of emicizumab is estimated to be 10% to 20% of normal.29 Although effective, these therapies require repeated administration and, thus far, incompletely prevent joint disease.30,31 This review will focus on the clinical development of Roctavian for which the collective clinical trials3-7 (Table 1) represent the largest body of work thus far available for any licensed AAV product. From this important proof-of-concept work, several outstanding questions have emerged.14,32 Most notably, unexplained year-over-year declines in FVIII expression after Roctavian treatment that contrast with observed stable FVIII expression in AAV HA gene therapy clinical trials with more modest FVIII expression (Figure 1; supplemental Table 1)12,13 and observations in hemophilia B gene therapy efforts.2,33,34 Available data are important to carefully consider and outline when consenting patients as well as to mechanistically inform the rational design of next-generation HA gene therapy approaches.
Name . | Phase . | Dose (vg/kg) . | Unique criteria . | N . | Immunosuppression . | Outcomes . | Trial status . | NCT# . | References . |
---|---|---|---|---|---|---|---|---|---|
BMN270-201 | 1/2 | 6 × 1012 | See text | 1 | None | See text | Active, not recruiting | 02576795 | 3,4,7 |
2 × 1013 | 1 | None | |||||||
6 × 1013 | 7 | 7/7 (prophylactic) | |||||||
4 × 1013 | 6 | 4/6 on-demand | |||||||
BMN270-203 | 1/2 | 6 × 1013 | Positive AAV Ab | 1 | - | No available data | Active, not recruiting | 03520712 | - |
BMN270-205 (GENEr8-INH) | 1/2 | 6 × 1013 | Active or prior FVIII Inhibitors | 4 | 4/4 | N = 2 active inhibitors, 1 amnestic response, 1 sign of tolerance; N = 2 prior inhibitors and did not develop inhibitors | Recruiting | 04684940 | 35 |
GENEr8-1 | 3 | 6 × 1013 | See text | 134 | 106/134 | See text | Active, not recruiting | 03370913 | 5,6 |
GENEr8-2 | 3 | 4 × 1013 | See text | 1 | 1/1 | See text | Completed | 03392974 | 36 |
GENEr8-3 | 3 | 6 × 1013 | Prophylactic steroids | 22 | 40 mg per day steroids for 8 weeks; taper dose for 11 weeks | >90% participants experienced ALT elevations; FVIII 14.5% at year-1 | Active, not recruiting | 04323098 | 37 |
Name . | Phase . | Dose (vg/kg) . | Unique criteria . | N . | Immunosuppression . | Outcomes . | Trial status . | NCT# . | References . |
---|---|---|---|---|---|---|---|---|---|
BMN270-201 | 1/2 | 6 × 1012 | See text | 1 | None | See text | Active, not recruiting | 02576795 | 3,4,7 |
2 × 1013 | 1 | None | |||||||
6 × 1013 | 7 | 7/7 (prophylactic) | |||||||
4 × 1013 | 6 | 4/6 on-demand | |||||||
BMN270-203 | 1/2 | 6 × 1013 | Positive AAV Ab | 1 | - | No available data | Active, not recruiting | 03520712 | - |
BMN270-205 (GENEr8-INH) | 1/2 | 6 × 1013 | Active or prior FVIII Inhibitors | 4 | 4/4 | N = 2 active inhibitors, 1 amnestic response, 1 sign of tolerance; N = 2 prior inhibitors and did not develop inhibitors | Recruiting | 04684940 | 35 |
GENEr8-1 | 3 | 6 × 1013 | See text | 134 | 106/134 | See text | Active, not recruiting | 03370913 | 5,6 |
GENEr8-2 | 3 | 4 × 1013 | See text | 1 | 1/1 | See text | Completed | 03392974 | 36 |
GENEr8-3 | 3 | 6 × 1013 | Prophylactic steroids | 22 | 40 mg per day steroids for 8 weeks; taper dose for 11 weeks | >90% participants experienced ALT elevations; FVIII 14.5% at year-1 | Active, not recruiting | 04323098 | 37 |
Data are current as of March 2024.
Ab, antibody.
The vector
Roctavian is a recombinant AAV5 vector consisting of a single-stranded DNA cassette comprised of a hepatocyte-specific promoter, encoding a codon-optimized B-domain–deleted FVIII variant, FVIII-SQ.3 The ∼5 kilobase (kb) expression cassette (including the inverted terminal repeats),38 like all known HA vectors, exceeds the 4.7-kb wild-type AAV genome and approximate packaging capacity of AAV vectors.39 Roctavian is manufactured using the baculovirus Spodoptera frugiperda (Sf9) insect cell production system. AAV vectors manufactured with the baculovirus-Sf9 platform have more trunctated genomes and genomic heterogeneity,40,41 which are exacerbated by oversized cassettes,42 as well as reduced potency relative to vectors manufactured in mammalian cells.43,44
Phase 1/2 clinical trial
Enrollment criteria and dose escalation
The phase 1/2 study of Roctavian, then BMN270, enrolled adult male participants with severe HA with ≥150 FVIII exposures, no history of FVIII inhibitors, and without detectable anti-AAV5 neutralizing antibodies (NAb), significant liver disease, or active hepatitis B or C.3 Additionally, participants were either on FVIII prophylaxis or, if treated on demand, had ≥12 bleeding episodes in the 12 months before study enrollment. Roctavian demonstrated a clear, albeit steep, dose response between the 4 AAV vector doses studied (6 × 1012, 2 × 1013, 4 × 1013, and 6 × 1013 vector genomes per kg body weight [vg/kg]) and FVIII levels at year 1 (Figure 2).
Safety observations and glucocorticoid use
All 7 participants in the highest-dose cohort, now the licensed dose (6 × 1013 vg/kg), received immunosuppression with glucocorticoids after the first participant in the dose cohort developed an elevated alanine aminotransferase (ALT) of 1.5-fold his baseline. There was no definitive temporal association between elevated ALT and decreased FVIII levels and all transaminase elevations were asymptomatic. Treatment-related serious adverse events of interest also included 1 infusion reaction in a participant in the 4 × 1013 vg/kg–dose cohort who was hospitalized after the development of fever, headache, and myalgia within 24 hours of vector administration. There were no incidences of thrombosis or inhibitor development.3,4
Preliminary efficacy
Consistent with FVIII expression in the range of, or below, moderate HA, participants in the 2 lowest dosing cohorts (6 × 1012 and 2 × 1013 vg/kg) demonstrated no improvement in bleeding. Participants in the 2 highest dosing cohorts (4 × 1013 to 6 × 1013 vg/kg) initially achieved FVIII levels in either the range of mild HA or normal FVIII such that their annualized bleeding rate (ABR) at year 1 decreased from a median of 16 bleeds to 0 bleeds per year, with a range from 0 to 5.4 bleeds per year.3,4 Subsequent longitudinal follow-up of phase 1/2 trial participants observed multiyear declines in FVIII expression (Figure 1). Importantly, despite declining FVIII activity, the median ABR of the 6 × 1013 vg/kg–dose cohort (n = 6) remained 0 bleeds per year through year 5 (mean ABR ranged from 0.7 to 1.3 bleeds per year).7
Pivotal trials Design
Phase 3 enrollment criteria and study population
Initially, 2 phase 3 studies for Roctavian were pursued, GENEr8-1 and GENEr8-2, which evaluated valoctocogene roxaparvovec at doses of 6 × 1013 vg/kg and 4 × 1013 vg/kg, respectively. The GENEr8-2 study enrolled a single participant and is now closed. Roctavian was licensed on the results of the GENEr8-1 pivotal trial (ClinicalTrials.gov identifier: NCT 03370913)5,6 that enrolled 134 participants who were aged, on average, 32 ± 10 years (range, 18-70). Notably, this cohort likely underrepresents viral comorbidities of patients with severe HA aged >50 years in which 90% contracted hepatitis C virus and 30% to 60% contracted HIV infection.45 Enrollment criteria were modified from phase 1/2 criteria3 to exclude participants with a prior liver biopsy demonstrating liver fibrosis of stage ≥3, liver cirrhosis of any etiology determined by liver ultrasound, a history of venous or arterial thrombosis, or a known thrombophilia.5,6 During phase 3 investigation, enrollment criteria were amended to exclude patients with HIV. This was in response to observed liver toxicities with the single participant in GENEr8-2 who was HIV positive and developed transaminase elevations, detailed below. Of 181 potential participants, 26 (14%) were ineligible because of preexisting anti-AAV5 antibodies assayed by a total binding antibody assay,6 whereas the AAV5 seroprevelance in a larger multinational cohort determined by Biomarin was 35%.46 The latter incidence of AAV5 antibodies is analogous to that reported among participants enrolled in the pivotal trial of etranacogene dezaparvovec (Hemgenix; CSL) determined by a functional, transduction based assay.47 The role of anti-AAV5 antibodies on Hemgenix efficacy will be determined in a postmarketing study.47 A separate trial of Roctavian that enrolled participants with positive anti-AAV5 antibodies has not yet publicly presented results (ClinicalTrials.gov identifier: NCT03520712; Table 1).
Clinical trial end points
Accelerated regulatory approval was initially pursued using a surrogate end point of median chromogenic assay (CSA)–determined FVIII activity at 23 to 26 weeks after vector administration. Based on this, an original Biologics Licensing Application was submitted to the US Food and Drug Administration (FDA) in December 2019. However, the agency determined that CSA FVIII activity did not meaningfully predict ABR, and accelerated approval was not granted.48 Instead, the FDA requested the sponsor to provide ≥2 years of follow-up data, including 1 year of follow-up off immunomodulation with a revised primary end point of ABR. Roctavian was FDA approved in June 2023 on a primary outcome of ABR noninferiority before and after infusion.48
Measurement of transgene biologic activity by FVIII assay
Roctavian transgene-derived FVIII activity measures ∼1.6-fold lower by CSA than 1-stage assay (OSA),3 which was subsequently replicated by multiple HA AAV gene transfer trials12,13,15 targeting hepatocyte expression of FVIII. This observation was unexpected because OSA assay determined FVIII activity with recombinant FVIII-SQ protein (eg, Xyntha and Refacto) has either been the same or lower than CSA-determined FVIII activity.49 Although a mechanistic explanation is unknown, it is reproducible across trials (supplemental Table 2).3,12,15 This suggests a common biological mechanism that may be secondary to FVIII expression in hepatocytes and/or codon optimization F8 complementary DNA that affects functional properties of the translated protein, the latter previously demonstrated for FIX.50 OSA FVIII activity captures the generation of the activated FVIII (FVIIIa) cofactor from FVIII and downstream amplification to fibrin formation, whereas the CSA measures FVIIIa contribution to the activation of FX only.49 Biochemical studies suggest that transgene-derived FVIII may have enhanced kinetics of activation or function51,52 that would be captured by OSA but not CSA, suggesting that the OSA may better predict clinical phenotype. Consistent with this, OSA FVIII activity correlated with participant phenotype after Roctavian6 and dirloctocogene samoparvovec.12
Phase 3 safety
Infusion reactions, immune response to FVIII, prothrombotic risk, and creatine kinase elevations
Infusion reactions occurred in 5% of GENEr8-1 participants that were managed by slowing or pausing the rate of infusion and other supportive measures that included antihistamines, antipyretics, and glucocorticoids.5 This frequency is analogous to that observed with Hemgenix, also an AAV5 vector administered at a similar dose.20,47 All participants completed Roctavian administration. Additionally, the FDA summary for regulatory action of Roctavian outlined that 4 participants had a single, nonsustained positive Bethesda result (>0.6 Bethesda units).48 Ten participants tested positive for anti-FVIII total binding antibodies that did not correlate with FVIII activity.48 Sixty-five percent of participants had positive peripheral blood mononuclear cell reactivity to FVIII peptides by interferon-γ enzyme-linked immunospot assay that also did not correlate with FVIII activity.48 Overall, that these immune responses to FVIII did not correlate with phenotype is reassuring.
Some participants had transiently elevated FVIII activity. Although venous thromboembolism (VTE) was not observed, elevated FVIII has been observed in at least 2 trials,5,15 raising the possibility of VTE after HA gene transfer. Elevated FVIII activity is an independent, dose-dependent risk factor for VTE.53-62 For example, FVIII activity of >250% has an VTE odds ratio of 20.53 A blanket strategy for management of prothrombotic risk in the setting of supratherapeutic FVIII expression is unclear and the risks and benefits of long-term anticoagulation must be carefully considered. However, the authors may consider prophylactic anticoagulation in patients with sustained FVIII expression of >250%, particularly in the setting of other concurrent prothrombotic risk factors and/or a prior history of thrombosis.
Lastly, nearly half of pivotal trial participants experienced creatine phosphokinase (CPK) elevations of unknown source.63 Roctavian mouse preclinical pharmacology-toxicology studies demonstrated histopathological findings of cardiac fibrosis and inflammation at doses of 6 × 1013 vg/kg that were not reported in nonhuman primate toxicology studies,48 making the translational relevance of mouse observations unclear. Creatine kinase elevations were also observed with Hemgenix,47 an AAV5 vector produced in Sf9 cells administered at a comparable dose. CPK elevations were not reported in other AAV hemophilia trials,1,12,33,64 although routine measurements were not done. It is unclear whether CPK observations represent transduced tissue (eg, cardiac or skeletal muscle cells) toxicity or whether observations are AAV5 specific vs generalizable to the AAV vector-dose and/or manufacturing platform.
Transaminase elevations and glucocorticoid use
Among all participants on the GENEr8-1 trial (n = 134), 86% had adverse event qualifying ALT elevations. Although most ALT elevations were modest, elevations of >5- to 20-fold the upper limit of normal were reported in a small number of patients.5 On average, elevations occurred 8 weeks after Roctavian administration and most occurred within 3 months after vector, but transaminase elevations at multiple months to 2 years after vector were observed in 29% of participants.5,6,48 The etiology of these late rises in transaminases is unclear. Glucocorticoids were used in 106 (79%) participants and initiated at a median of 8 weeks after vector because of transaminase elevations.5,6,65 Two participants required IV methylprednisolone because of the magnitude of the ALT increases (247 and 448 U/L) without evidence of hepatic synthetic dysfunction.5 The median duration of glucocorticoid use was 230 days (range, 22-551).5,6,48 Overall, 80% of participants that used glucocorticoids experienced side effects such as acne, insomnia, Cushing syndrome, and weight increase5,6; 3% of participants experienced serious adverse events with glucocorticoids use, including steroid-induced diabetes and hypertension.5 Tacrolimus and mycophenolate mofetil were used as steroid-sparing immunosuppressants in 27 participants.5,6 Roctavian prescribing information recommends monitoring ALT weekly for 26 weeks after vector infusion and initiating corticosteroids in response to ALT elevations of ≥1.5× baseline and continuation until ALT returned to baseline. As such, it is prudent for prescribers to have at least 2 ALT measurements before vector to determine the patients’ baseline. Although not definitive, the mechanism of ALT elevation was not deemed by the investigators to be associated with a cellular immune response to the AAV capsid observed in other work4,5,66 and remains unclear; however, most initial ALT elevations were observed in the expected window of an anti-AAV capsid immune response. Additionally, Roctavian prescribing information notes that ALT elevation was acutely associated with a 30% decline in FVIII activity,67 which may be consistent with a capsid immune response.
Transaminase elevations and rationale to exclude patients who are HIV positive
Transaminase elevations in a small number of patients who are HIV positive resulted in amended enrollment criteria to exclude patients who are HIV positive. Three participants who are HIV positive received Roctavian, including the only patient on GENEr8-2 trial. The singl participant on the GENEr8-2 trial experienced a rapid in ALT, beginning day 34 after Roctavian infusion. Despite oral prednisone, his ALT rose, peaking at about 800 IU/L, and did not plateau until the nonnucleoside reverse transcriptase inhibitor (efavirenz) of his highly active antiretroviral therapy was paused on day 52 after vector. Per patient request, efavirenz was restarted but was associated with further increases in liver enzymes by day 63. Subsequently, an alternative HIV treatment regimen was initiated, and liver enzymes improved. He had no measurable FVIII activity after week 5.36 Subsequent studies conducted in primary human hepatocytes demonstrated that efavirenz dose-dependently decreased FVIII transcription in vitro and F8 messenger RNA levels were not restored after discontinuation.48 Although a mechanistic basis for clinical observations is unclear, because of the experience with this subject, GENEr8-2 was suspended and patients who are HIV-positive were excluded from GENEr8-1.
Vector shedding
Adeno-associated viral vectors are shed in bodily fluids for a duration that directly correlates with vector dose and varies by AAV capsid. Roctavian vector DNA was quantified by quantitative polymerase chain reaction (PCR) of a FVIII-SQ–specific amplicon until 3 consecutive negative samples were achieved in participants of phase 1 to 3 trials (n = 149) in the urine, saliva, semen, and feces, as well as in the blood to assess vector shedding and biodistribution, respectively. Vector amplicons were detected at the first postinfusion time point, day 2 after vector, and median peak vector DNA was found in all samples typically within 2 weeks after infusion.3,4,68 At the licensed dose of 6 × 1013 vg/kg, the median time to vector concentration below the limit of detection in semen was 20 weeks (maximum, 36 weeks).4 Prolonged shedding was observed in feces, with median time to below the limit of detection of 120 weeks,4 which is hypothesized to be because of shedding of transduced gut epithelial cells or peripheral blood mononuclear cells; this is consistent with persistent detection of vector sequences in whole-blood samples.4 PCR alone cannot distinguish whether vector DNA is free or encapsidated in a vector particle. Immunocapture PCR was used to quantify encapsidated vector DNA in participant plasma and semen and found that intact particles were present for a maximum of 12 weeks in both fluids,69,70 which could theoretically transduce cells upon exposure.
The presence of intact vector in bodily fluids after infusion raises the possibility that untreated individuals may be exposed to vector. Importantly, wild-type AAV is nonpathogenic and all viral genes are removed from AAV vectors, rendering them replication incompetent.71 Direct investigation of this in people and animal models is limited. Analysis of nonhuman primates’ bodily fluids after intramuscular AAV administration demonstrated that infectious particles were only recovered from the serum within the first 72 hours after vector administration, which differed from the presence of a transient PCR signal in most body fluids much beyond 72 hours.72 Similarly, functional AAV was identified for up to 48 hours in sheep stool and 4 days in rabbit semen73 after IV AAV delivery. Although it is not clear how long functional AAV is shed after administration, data support that it is on the order of days. Nonetheless, an additional concern is that persistent AAV vector shedding, although not infectious, may incite a sufficient immune response to result in AAV seroconversion (ie, the development of anti-AAV NAb) in close contacts. Importantly, there is no documentation of this in humans, but it has not been extensively studied. Studies in mice demonstrated that topical optical delivery (intended to mimic exposure via respiratory droplets from a person shedding AAV) at vector doses analogous to clinical vector shedding values did not induce seroconversion in cage contacts.74 Although limited, animal models suggest that exposure to AAV vector doses analogous to what may be encountered from close contact with a patient after AAV are insufficient to result in AAV antibody formation; thus, the risk of close contact/healthcare worker seroconversion is unlikely. In the absence of definitive information, the Children’s Hospital of Philadelphia (CHOP) developed institutional AAV vector shedding guidelines for clinical care and close patient contacts (supplemental Table 3). A concurrent study to evaluate seroconversion of household contacts of patients treated AAV is ongoing at CHOP. Lastly, although federal guidance for AAV gene therapy biosafety in clinical settings does not yet exist, the Centers for Disease Control and National Institutes of Health Guidelines classify AAV vectors as risk group 1 agents that can generally be handled at biosafety level 1 containment, pending the vector does not encode a toxic or integrating transgene.8
Integration and oncogenicity risk
Although AAV is predominantly nonintegrating, random integration of environmental wild-type AAV and recombinant AAV vectors has been demonstrated in animal models and clinical studies without evidence of tumorigenesis and genotoxicity in humans. Nonetheless, this raises the theoretical risk of genotoxicity that could progress to malignancy. Although malignancies have been observed after AAV vector administration (supplemental Table 4), including 2 malignancies after Roctavian, these events have thus far not been linked to AAV. Risk of insertional mutagenesis after AAV has been extensively studied and is comprehensively reviewed elsewhere.75
Phase 3 efficacy
Factor VIII expression and ABR data
Direct data from the FDA’s summary basis for regulatory action of Roctavian define median and mean OSA and CSA FVIII activity for phase 3 trial participants at 12-, 24- and 36-months after vector administration (Table 2).48 Given FVIII expression data are not normally distributed, we prefer discussing median values as the most representative measure of data central tendancy.32,48 Continued multiyear declines in FVIII expression have been reported out to 6 years in phase 1/2 participants,4,7,48 which is similarly observed in phase 3 participants (Figure 1; Table 2) albeit with lesser follow-up duration. Notably, peak FVIII values were ∼2-fold higher for phase 1/2 vs phase 3 participants. Nonetheless, across phase 1-3 participants, year-2 FVIII activity declined by ∼40% without apparent subsequent plateau expression,6,32,48 making phenotypic classification dependent on the duration of follow-up after vector administration (Figure 3). Although the minimal threshold of therapeutic FVIII activity is undecided, information from patients with mild HA demonstrates that FVIII activity of approximately 20% of normal protects against spontaneous and/or all joint bleeds,24 which is also consistent with the estimated FVIII equivalency of emicizumab prophylaxis.29 Together, these data may support that durable FVIII activity of >20% is a minimal therapeutic threshold for HA gene therapy to improve upon current standards of care and eliminate joint bleeds.
Time after vector (months) . | Rollover population N = 112 . | Directly enrolled population N = 22 . | ||
---|---|---|---|---|
CSA . | OSA . | CSA . | OSA . | |
Month 12 | N = 111 | N = 111 | N = 21 | N = 21 |
Mean (SD) | 43.6 (45.5) | 64.7 (64.6) | 38.2 (46.3) | 59.7 (67.0) |
Median (Q1, Q3) | 24 (12.5, 63.7) | 40.0 (20.4, 87.5) | 23.9 (11.2, 52.8) | 40.5 (17.4, 82.6) |
Min, max | 0, 231.2 | 0, 311.11 | 1.6, 207.4 | 4.4, 294.1 |
Month 24 | N = 98 | N = 99 | N = 19 | N = 18 |
Mean (SD) | 25.0 (35.5) | 38.9 (50.7) | 22.0 (28.7) | 36.0 (40.8) |
Median (Q1, Q3) | 12.7 (5.1, 26.5) | 22.7 (7.9, 45.7) | 8.9 (5.8, 25.9) | 19.5 (7.9, 37.7) |
Min, max | 0, 187.1 | 0, 271.3 | 0, 110.6 | 2.4, 146.7 |
Month 36 | N = 96 | N = 97 | N = 15 | N = 15 |
Mean (SD) | 21.0 (34.0) | 33.8 (47.6) | 20.8 (24.4) | 32.2 (33.1) |
Median (Q1, Q3) | 10.0 (4.3, 19.8) | 17.1 (7.2, 35.1) | 9.4 (6.6, 31.7) | 20.6 (8.5, 46.7) |
Min, Max | 0, 217.7 | 0, 291.4 | 0, 74.5 | 1.9, 104.2 |
Time after vector (months) . | Rollover population N = 112 . | Directly enrolled population N = 22 . | ||
---|---|---|---|---|
CSA . | OSA . | CSA . | OSA . | |
Month 12 | N = 111 | N = 111 | N = 21 | N = 21 |
Mean (SD) | 43.6 (45.5) | 64.7 (64.6) | 38.2 (46.3) | 59.7 (67.0) |
Median (Q1, Q3) | 24 (12.5, 63.7) | 40.0 (20.4, 87.5) | 23.9 (11.2, 52.8) | 40.5 (17.4, 82.6) |
Min, max | 0, 231.2 | 0, 311.11 | 1.6, 207.4 | 4.4, 294.1 |
Month 24 | N = 98 | N = 99 | N = 19 | N = 18 |
Mean (SD) | 25.0 (35.5) | 38.9 (50.7) | 22.0 (28.7) | 36.0 (40.8) |
Median (Q1, Q3) | 12.7 (5.1, 26.5) | 22.7 (7.9, 45.7) | 8.9 (5.8, 25.9) | 19.5 (7.9, 37.7) |
Min, max | 0, 187.1 | 0, 271.3 | 0, 110.6 | 2.4, 146.7 |
Month 36 | N = 96 | N = 97 | N = 15 | N = 15 |
Mean (SD) | 21.0 (34.0) | 33.8 (47.6) | 20.8 (24.4) | 32.2 (33.1) |
Median (Q1, Q3) | 10.0 (4.3, 19.8) | 17.1 (7.2, 35.1) | 9.4 (6.6, 31.7) | 20.6 (8.5, 46.7) |
Min, Max | 0, 217.7 | 0, 291.4 | 0, 74.5 | 1.9, 104.2 |
FVIII activity levels in IU/dL; data are from the FDA summary basis for regulatory action of valoctocogene raxaparvovec.48
Min, minimum; max, maximum; SD, standard deviation.
Analysis of ABR before and after Roctavian was prospectively assessed in the 112 roll-over participants for whom a 6-month prospective noninterventional study (270-902) of standard-of-care FVIII prophylaxis was conducted and compared with ABR after vector administration. An additional 22 patients were enrolled directly into GENEr8-1. From the FDA’s summary basis for regulatory action of Roctavian, the roll-over population demonstrated an 84% reduction in mean ABR such that the pre-vector mean ABR of 5.4 was reduced to 2.6. Of the 112 roll-over population, 13 (12%) patients returned to prophylaxis at a median of 2.3 years (range, 0.1-3.3) after vector administration for which an ABR of 35 was imputed for the analysis.48 Thus, one may anticipate that the ABR after vector is overestimated for an individual recipient. Among available longitudinal follow-up of the entire 134 participant cohort, 75% had 0 bleeds through year 1, whereas 10% (n = 13) had increased bleeding after Roctavian.5
It is notable that the initiation of this pivotal trial preceded widespread adoption of emicizumab for prophylaxis. Although direct comparison of Roctavian and emicizumab have not been undertaken, Biomarin underwrote a crosstrial comparison between GENEr8-1 (n = 132) and HAVEN 3 Group D (n = 63). Matching-adjusted indirect comparison (MAIC) was used to correct for statistically significant differences in age, bleeding before enrollment, and type of FVIII products used before intervention between the 2 studies.76 Participants in Group D of HAVEN3 were on FVIII prophylaxis before initiating weekly emicizumab,26 and thus, were most like GENEr8-1 participants. The unadjusted ABR for both Group D and GENEr8-1 participants while on FVIII prophylaxis was 4.8 bleeds per year.5,26 In the MAIC-weighted crosstrial comparison, the mean ABR for Group D participants on emicizumab (3.3 bleeds per year) was lower than that for GENEr8-1 participants while on FVIII prophylaxis during the lead-in noninterventional study (5.5 bleeds per year), with a rate ratio of 1.53 (95% confidence interval, 1.00-2.23). However, the mean ABR for Group D was higher than for GENEr8-1 participants 4 to 52 weeks after receiving Roctavian (1.82 bleeds per year), with a rate ratio of 0.55 (95% confidence interval, 0.33-0.93). Consistent with prior results, emicizumab provided better hemostatic prophylaxis than FVIII replacement,26 but Roctavian appears to provide better prophylaxis during year 1 according to the MAIC model that one may anticipate would not be maintained longitudinally given declining FVIII expression after Roctavian over time. Additional information of how Roctavian efficacy compares with current standards of HA care, including emicizumab, may emerge from an outcome-based agreement with the German healthcare system in which Roctavian reimbursement will be established based on real-word data collected from the German Haemophilia Registry collected over the next 3 years.77
Drug interactions and expression
One participant on the phase 3 trial (ClinicalTrials.gov identifier: NCT03370913) experienced a decline in FVIII activity (from 75 to <3 IU/dL) at week 64 after vector administration after initiating isotretinoin for steroid-induced acne that only returned to approximately half of preisotretinoin values after discontinuation.5 In vitro mechanistic studies using primary human hepatocytes demonstrated reduced FVIII RNA, but not DNA, when cultured with isotretinoin, suggesting reduced FVIII expression that, like clinical trial observations, was only partially restored after removing isotretinoin.78
Expression durability
As a general observation in the field, HA AAV vectors at doses that initially achieved normal/near-normal FVIII expression observed unexpected declines in expression.6,15,48 In contrast, at least 4 AAV-based gene therapy trials for HA (ClinicalTrials.gov identifiers: NCT03588299, NCT03003533, NCT03061201, and NCT03001830) demonstrated 2 to 5 years of stable FVIII expression in the lower half of mild HA or moderate HA.12,13,15,79 Consistent with human observations, stable FVIII expression in the range of mild HA was demonstrated in HA dogs for up to 10 years.80,81 Furthermore, we recently reported that long-term durability of FVIII expression in HA mice after AAV vector produced from mammalian cells was expression-level dependent, that is, high levels of FVIII expression were not durable whereas lower levels were.82 Thus, the combined preclinical and clinical studies suggest that AAV gene addition can impart durable FVIII expression, although possibly only in the range of moderate or mild HA.
The mechanism for declines in FVIII expression after AAV-mediated gene transfer are being investigated by multiple groups. Although prior work in mice concluded that capsid- or manufacturing-platform–specific epigenetic changes to expression are responsible for modifying transgene expression,42,83 these explanations are inconsistent with other observations. First, AAV5 has demonstrated multiyear stable FIX-Padua expression,14,20 suggesting that observations are not AAV5 specific. Second, loss of initial normal/near-normal FVIII expression has been demonstrated with Sf9-manufactured AAV5 and AAV6 for which the latter vector also achieved multiyear durable FVIII expression in the range of low mild or moderate HA, suggesting that durable FVIII expression is possible from vector manufactured in Sf9 cells. At a minimum, this would make epigenetic modifications or gene silencing dependent on the level of FVIII expression or vector dose. Last, our data in mice treated with the same lot of mammalian cell (HEK293)-produced AAV8 vector from an oversized cassette (5 kb) demonstrated stable low level but not high level FVIII expression in mice,82 which again suggests against loss of FVIII expression being specific to the Sf9 manufacturing platform.
Additionally, given that the size of FVIII-SQ complementary DNA typically requires that FVIII vector cassettes exceed the 4.7-kb wild-type AAV genome raises the question that incomplete cassette packaging and virion heterogeneity, exacerbated by the Sf9 platform,40,42,43 may preclude stable FVIII expression. Inconsistent with this hypothesis, the cassettes, inclusive of the inverted terminal repeats, of Roctavian3 and dirloctocogene samoparvovec12 are both 5 kb and dirloctocogene samoparvovec has demonstrated multiyear, approximately stable FVIII expression in the range of moderate/mild HA.12 Furthermore, declines in FVIII expression after Roctavian4,6,7 and in animal studies82 occurred multiple months and years after vector administration and outside the timing of stable episome formation, including that demonstrated with Roctavian in mice and nonhuman primates.84 Thus available clinical and animal data support that a 5-kb cassette is capable of stable episome formation, although virion heterogeneity is likely.85 In contrast to mechanisms implicating the vector and/or cassette properties, decades of prior experience manufacturing recombinant FVIII in mammalian cell culture identified a FVIII dose-dependent, positive relationship between FVIII expression and endoplasmic reticulum stress response.86-88 Furthermore, FVIII dose-dependent induction of endoplasmic reticulum stress in mice after AAV gene transfer is similarly supported.65,82,88-90 Thus, although not definitive, available data do not support that the vector construct, capsid, manufacturing platform, or gene silencing are the only mechanism responsible for loss of FVIII activity when initial near normal/normal FVIII levels are achieved after AAV-mediated gene transfer. Instead, available clinical trial data and observations of longitudinal FVIII expression in mice82 suggest that there are limitations on the level of tolerated durable FVIII expression. If this is ultimately correct, future efforts may substantially benefit from a gain-of-function FVIII variant18,19,82,91 that is able to normalize FVIII hemostatic function at levels of expression that permit durability.
AAV neutralizing antibodies
Although the implications of preexisting anti-AAV antibodies are incompletely understood, data support that high-titer AAV antibodies preclude systemic AAV vector administration efficacy by interfering with transduction via disrupting cell surface binding and/or result in reticuloendothelial AAV clearance.92-95 As expected after systemic AAV vector administration, participants who received Roctavian96 universally resulted in high-titer, multiserotype cross-reactive AAV antibodies of ≥1:1000.1,33 Data from the first cohort of patients to receive a systemic AAV vector demonstrated that AAV antibodies persist up to 15 years after AAV,97 which is generally consistent with lifelong immunity from environmental AAV exposure. Lastly, to date, there is no clinically proven strategy to overcome high titer AAV antibodies. Thus, with the current state of clinical development, patients are likely to only be eligible to receive AAV gene therapy once. The development of anti-AAV antibodies underscores the critical role of durability and the significance of the decision choosing to receive an AAV vector.
Procedures and informed consent considerations
Unlike cell-based gene therapy (eg, chimeric antigen receptor T cells for leukemia, or exagamglogene autotemcel for sickle cell anemia) that is regulated by the Foundation for Accreditation of Cellular Therapy, there is no regulatory body that oversees AAV vector administration and, thus far, no AAV vector is required to be part of the Risk, Evaluation, and Mitigation Strategy program. Thus, there is significant responsibility on the consenting physician and patient to ensure risks, benefits, and unknowns are carefully considered in the informed consent process (Table 3), which we have outlined in an institutional informed consent document for Roctavian at CHOP (supplemental Data). Additionally, although there is no required long-term monitoring by regulatory agencies, the World Federation of Hemophilia established a long-term follow-up registry98 that aims to capture all patients with hemophilia globally that receive gene therapy to permit evaluation of possible unexpected and/or rare events.98
|
• General knowledge of the rationale and approach |
• General knowledge about AAV vectors |
|
• Short-term safety: infusion reactions, transaminase, and CPK elevations |
• Likelihood/plan for immune modulation and glucocorticoid side effects |
• Long-term safety considerations: unclear implications on liver health and theoretical risk of genotoxicity |
|
• Expression profile and durability |
• Range of phenotypic improvement at 1 year vs longitudinally; frequency of therapeutic response vs nonresponse |
• Development of anti-AAV neutralizing antibodies and anticipated inability to receive future AAV therapies unless there are scientific developments to overcome anti-AAV neutralizing antibodies. |
|
• Infusion day |
• Anticipated short-term follow-up and monitoring; baseline liver ultrasound and blood tests |
• Anticipated long-term follow-up and monitoring, including recommendations around long-term liver health and minimizing potential liver toxicities (eg, alcohol use, nonalcoholic fatty liver disease); planned long-term liver health evaluations and durability of expression |
• Vector shedding precautions for clinic visits and close contacts (see supplemental Table 1) as well as contraindications for blood and organ donation after valoctocogene roxaparvovec |
|
• General knowledge of the rationale and approach |
• General knowledge about AAV vectors |
|
• Short-term safety: infusion reactions, transaminase, and CPK elevations |
• Likelihood/plan for immune modulation and glucocorticoid side effects |
• Long-term safety considerations: unclear implications on liver health and theoretical risk of genotoxicity |
|
• Expression profile and durability |
• Range of phenotypic improvement at 1 year vs longitudinally; frequency of therapeutic response vs nonresponse |
• Development of anti-AAV neutralizing antibodies and anticipated inability to receive future AAV therapies unless there are scientific developments to overcome anti-AAV neutralizing antibodies. |
|
• Infusion day |
• Anticipated short-term follow-up and monitoring; baseline liver ultrasound and blood tests |
• Anticipated long-term follow-up and monitoring, including recommendations around long-term liver health and minimizing potential liver toxicities (eg, alcohol use, nonalcoholic fatty liver disease); planned long-term liver health evaluations and durability of expression |
• Vector shedding precautions for clinic visits and close contacts (see supplemental Table 1) as well as contraindications for blood and organ donation after valoctocogene roxaparvovec |
Future perspectives
The marked progress in AAV gene addition for HA and hemophilia B as well as other disorders suggests the possibility for a long-awaited therapeutic paradigm of “one-and-done” therapy for monogenic disease and continued progress developing gene-based medicines. Careful mechanistic explanation for observations that have emerged from the clinical development of Roctavian will undoubtedly inform next-generation products. It is the authors’ belief that the goal of HA gene therapy to impart sustained FVIII expression that normalizes hemostasis for all recipients will be achieved through an iterative process building on prior accomplishments. Although he development and approval of Roctavian is a landmark in HA therapeutics, next-generation approaches are needed for gene therapy for HA to fulfill its promise.
Authorship
Contribution: All authors contributed to writing the manuscript.
Funding acknowledgement: NIH/NHLBI K08 HL146991 (LA George)
Conflict-of-interest disclosure: L.A.G. is on the scientific advisory board of Form Bio; has been a consultant for Intellia, Biomarin, Pfizer, Spark Therapeutics, and Tome Biosciences; and receives royalty payments from AskBio Therapeutics. B.J.S.-J. is on the scientific advisory board of GeneVentiv and Amarna; has been a consultant for Genentech, Biomarin, and Pfizer; and has received royalty payments from Cabaletta. J.C.S. declares no competing financial interests.
Correspondence: Lindsey A. George, Department of Pediatrics at the University of Pennsylvania School of Medicine, Colket Translational Research Building, Room 5016, 3501 Civic Center Blvd, Philadelphia, PA 19104; email: [email protected].
References
Author notes
The full-text version of this article contains a data supplement.