Management of acute infusion-related reactions in AAV gene therapy guided by mechanistic insight

Abstract

Infusion-related reactions (IRRs), serious immune events that occur either during or within six hours of an infusion, can occur with adeno-associated virus (AAV) gene therapy. IRRs may be attributed to IgE-mediated hypersensitivity, leading to treatment discontinuation due to concern about progressive, life-threatening allergic reactions. IRRs were observed in two participants who received an AAV8 gene therapy (DTX401) for glycogen storage disease type Ia and in one participant who received an AAV9 gene therapy (UX701) for Wilson’s disease. All of these IRRs were associated with a rapid infusion rate, despite the second and third participants receiving antipyretic and antihistamine pre-medication. In both participants with DTX401-related IRRs, the infusion was terminated and not restarted. The DTX401 study protocol was amended to require both pre-medication and a slow, staged infusion; no additional IRRs were reported. Due to a medical error, a UX701-treated participant received a very rapid initial infusion rate, resulting in an IRR. After symptoms resolved, administration was restarted using a slow, staged infusion and was completed uneventfully. AAV gene therapy IRRs may be rate related and possibly driven by complement activation. In our studies, IRRs were mitigated by a slow, staged infusion.

Graphical abstract

Infusion-related reactions (IRRs) can occur with AAV gene therapy. IRRs were observed in two participants who received AAV8 gene therapy and in one who received AAV9 gene therapy. AAV gene therapy IRRs may be rate related, possibly driven by complement activation. In our studies, a slow, staged infusion mitigated additional IRRs.

Introduction

Infusion-related reactions (IRRs) are serious immune events with features of hypersensitivity that occur either during or within six hours of an infusion. IRRs have been observed in clinical trials of adeno-associated virus (AAV) gene therapies.^1,2,3,5 In some cases, the infusions were terminated.⁶ In other cases, the infusions were paused or slowed and completed successfully. IRRs vary in presentation and severity.⁷

Multiple mechanisms have been postulated to drive IRRs. For IRRs to protein-based therapeutics, IgE-mediated reactions are relatively rare, in contrast to reactions to food allergens, insect venom, and small molecules, most of which are assumed to be IgE-mediated. However, these events are sometimes both called anaphylaxis.

Anaphylaxis, a word coined in the early 20^th century, combines the α- prefix meaning “not” and φύλαξις (phylaxis) meaning “protective”. Richet and colleagues observed that repeat administration of low doses of Physalia physalis extracts (commonly known as the man o' war or bluebottle) were deadly in pre-clinical models, in contrast to Pasteur’s rabies vaccine that offered protection.^8,9,10,11

In the clinical realm, the 2020 World Allergy Organization (WAO) guidelines cite five definitions of anaphylaxis that are based on kinetics, severity, and clinical symptoms but do not include biomarkers.¹² In the realm of academic immunology, anaphylaxis is defined as the most severe of immediate (type I) hypersensitivity reactions, is IgE-mediated, and requires prior exposure.¹³ The two definitions of anaphylaxis together may suggest that all clinical reaction symptoms are driven by IgE. However, IgE-mediated and complement-mediated reactions have overlapping clinical features, and both may meet current diagnostic criteria for anaphylaxis. There is little mechanistic information about IRRs specifically during AAV gene therapy administration.¹⁴

Herein, we describe three clinical trial participants with no prior exposure to AAV that developed IRRs to AAV gene therapies. In the two cases where tryptase was measured, there was no elevation of serum tryptase; one case demonstrated complement activation, and the other did not have a sample collected for complement testing. Neither of these two participants continued the infusions. The third participant with an IRR successfully completed dosing after pausing and slowing the infusion, suggesting that this was infusion rate-related rather than IgE-mediated.

We recommend that AAV gene therapy be administered with antipyretic and antihistamine pre-medication and by a slow, staged infusion (Table 1).¹⁵ If an IRR occurs during delivery of the AAV gene therapy, after symptoms resolve, resumption of the infusion at a slower rate may be attempted. Additionally, both complement and tryptase biomarkers should be collected for mechanistic insight (Table S1).

Table 1.

The slow, staged infusion schedule is conducted over approximately 3 h

Timing	People <40 kg	People ≥40 kg
0–60 min	3 mL/h	6 mL/h
61–80 min	8 mL/h	18 mL/h
81–100 min	15 mL/h	38 mL/h
101–120 min	30 mL/h	75 mL/h
121–180 mina	80 mL/h	200 mL/h

^aEnsure all contents of the infusion bag are infused, even if the infusion time exceeds 180 min and is within 24 h of dose preparation.

Methods

DTX401-CL301 GlucoGene clinical trial

DTX401 is an AAV8-based gene therapy designed to deliver a copy of the G6PC gene to hepatocytes, resulting in the production of normally functioning G6Pase. The formulation buffer of DTX401 contains 0.0030%–0.0325% (w/v) poloxamer 188.

The GlucoGene study, also referred to herein as DTX401-CL301, is an ongoing phase 3, randomized, double-blind, placebo-controlled study of AAV8-mediated gene transfer of G6PC in patients with glycogen storage disease type Ia (GSDIa) (NCT05139316).¹⁶ Patients with positive anti-AAV8 antibody tests at screening were excluded from the study. Participants were randomized 1:1 to receive blinded DTX401 (1.0 × 10¹³ GC/kg by droplet digital polymerase chain reaction [ddPCR]) or placebo. This is the same dose level as 6.0 × 10¹² GC/kg, by quantified PCR, that was evaluated in the phase 1/2 study.¹⁷ All fully dosed participants received a prophylactic blinded immunomodulatory regimen of prednisolone or placebo at day 15 which was maintained for 4 weeks and then slowly tapered. Participants randomized to placebo who continued with the study in year 2 received DTX401 at week 48 and are termed the crossover-DTX401 group. Following the first IRR, the informed consent was amended and investigators were advised to consider the administration of antihistamines prior to dosing, at their discretion; patients were also re-consented for the study. Following the second IRR, the study protocol was amended to change the infusion rate to a slow, staged infusion (Table 1) with a requirement for pre-medication with an H1 antagonist, an H2 antagonist, and acetaminophen/paracetamol. Placebo participants were re-screened for anti-AAV8 antibodies at week 36 to determine eligibility for DTX401 dosing at week 48, and those positive were excluded from dosing at week 48. Forty participants received DTX401 in this study.¹⁶

UX701-CL301 Cyprus²⁺ clinical trial

UX701 is a nonreplicating, recombinant AAV9 vector designed to deliver stable expression of the copper-transporting ATP7B transgene to hepatocytes via intravenous infusion to normalize copper trafficking and excretion from the body. The formulation buffer of UX701 contains 0.01% (w/v) poloxamer 188.

The Cyprus²⁺ study, also referred to herein as UX701-CL301, is an ongoing phase 1/2/3 study of UX701 gene therapy in adults with Wilson’s disease (NCT04884815). After seven participants were dosed, and after the DTX401 IRR cases, the pharmacy manual was updated to include the slow, staged infusion (Table 1). The modified infusion regimen has a very slow starting rate for the first hour, with gradual rate increases over the second hour. Pre-medication with an H1 antagonist, an H2 antagonist, and acetaminophen/paracetamol were also added.

The participant described here enrolled in the phase 1/2 open-label safety and dose-finding stage 1, received a dose of 1.0 × 10¹³ GC/kg, pre-medication with an H1 antagonist, an H2 antagonist, acetaminophen/paracetamol, and prophylactic oral corticosteroids on days 1–113.

Ethics and informed consent

Both studies presented here were approved by the Institutional Review Boards (IRBs) at participating centers and were conducted in accordance with the Declaration of Helsinki, International Council for Harmonization (ICH) Good Clinical Practice guidelines, and applicable local regulatory requirements. Written informed consent was obtained from all participants before any study-specific procedures were performed.

Anti-AAV8 capsid total antibody assay

The anti-AAV8 total antibody assay is a tiered electrochemiluminescent (ECL) sequential bridging immunoassay utilizing the MesoScale Discovery (MSD) platform. Unlabeled AAV8 capsid is used as the capture phase and ruthenylated (SulfoTagged) AAV8 capsid as detector. Following the sample incubation step, to allow binding of anti-AAV8 antibodies, the plate is washed to remove the unbound material, and SulfoTagged AAV8 capsid is added to detect the captured antibodies. Commercial anti-AAV8 mouse monoclonal antibody is used as a surrogate positive control in this assay. Tripropylamine substrate (TPA, MSD) is added to each well, and the plate is read on a MESO QuickPlex SQ 120 instrument.

Anti-AAV9 capsid total antibody assay

Similarly to the anti-AAV8 assay, the anti-AAV9 total antibody assay is an ECL sequential bridging immunoassay utilizing the MSD platform. Unlabeled AAV9 capsid is used as the capture phase and ruthenylated (SulfoTagged) AAV9 capsid as detector. Following the sample incubation step, to allow binding of anti-AAV9 antibodies, the plate is washed to remove the unbound material, and SulfoTagged AAV9 capsid is added to detect the captured antibodies. Commercial anti-AAV9 mouse monoclonal antibody is used as a surrogate positive control in this assay. TPA (MSD) is added to each well, and the plate is read on a MESO QuickPlex SQ 120 instrument.

Tryptase assay

Samples for tryptase from the site for participant A were measured at LabCorp using the ImmunoCAP Tryptase test (test code 83520); the healthy reference range is 2.2–13.2 μg/L. Samples for tryptase for participant B were measured with a reference range of 0–13.5 μg/L.

Poloxamer 188 SPT

At the time of evaluation, there was no validated skin prick testing (SPT) protocol reported in the literature. A multidisciplinary team including representatives from the Sponsor, the local allergist, and a research pharmacist agreed to test the following formulations of poloxamer 188: a 1× preparation of poloxamer 188 equal in concentration to that found in active drug, a 1:10 dilution, a 1:100 dilution, and 10× concentrated formulation in case of uneven distribution of poloxamer 188 in the infusion bag. In addition to the poloxamer 188 tests, SPTs for a positive control, histamine, and a negative control, saline, were also performed. At site A, SPTs were performed using bifurcated needles.

Anti-PEG total antibody assay and poloxamer 188 confirmation step

Total anti-PEG antibody was analyzed in pre-dose GlucoGene study serum samples using a direct colorimetric procedure that identified serum samples that contained detectable antibodies based on normalized optical density (OD) signals equal to or above the established screening cut point. Antibody detection was based on a multi-tiered approach consisting of screening (tier 1), confirmatory (tier 2), and titration (tier 3) assays. The confirmatory assay determined whether the signal in the screening assay was the result of specific antibodies by comparing the assay OD signals of samples in the presence or absence of excess free PEG or excess free poloxamer 188. The titration assay procedure measured anti-PEG titer in each sample by serial dilution of the confirmed positive samples. All cut points used for screening, confirmation, and titration assessment were established prior to sample analysis.

Anti-PEG IgE antibody assay

The ImmunoCAP platform is a solid-phase sandwich immunoassay that detects specific IgE antibodies in study samples. PEG was immobilized onto a cellulose solid phase to capture anti-PEG IgE antibodies in participant serum. For detection, an enzyme labeled antibody to IgE was added to form a complex. The addition of substrate following detection reagent incubation was used to react with the complex to produce a fluorescent signal. The fluorescent signal was directly proportional to the amount of anti-PEG antibody captured onto the solid phase and was used to compare to a standard curve to determine the concentration. The quantitative range of the assay was 0.1–100 kU_A/L.

Complement functional activity assays (CH50, AH50, and MBL)

Serum was tested in the Wieslab Complement System Screen ELISA (COMPL300, Svar Life Science, Malmo, Sweden) at McGill University Health Center (Montreal, Canada), following the manufacturer’s instructions. Briefly, the assay is based on the principle of measuring a neoepitope generated by the formation of the membrane attack complex (MAC also known as C5b-9) in vitro following specific stimulation.

For the CH50 assay (classical pathway), wells were pre-coated with IgM. For the AH50 assay (alternative pathway), wells were pre-coated with LPS. For the mannan-binding lectin (MBL) pathway, wells were pre-coated with mannan. Sera were diluted in assay-specific buffers¹⁸ and incubated for one hour at 37°C. After washing, alkaline phosphatase-conjugated anti-human C5b-9 was added and incubated at room temperature for 30 min. Additional washing was performed, and the alkaline phosphatase substrate was added. The wells were incubated for 30 min, and absorbance was read at 405 nm.

The amount of formed C5b-9 reflects the activity of each arm of the complement cascade. The result is expressed semi-quantitatively using the following formula:

$$Activity=100\%\times \left[mean{A}_{405}\left(sample\right)–mean{A}_{405}\left(negativecontrol\right)\right]/\left[mean{A}_{405}\left(positivecontrol\right)–mean{A}_{405}\left(negativecontrol\right)\right]\text{.}$$ (Equation 1)

Samples and standards are typically measured in duplicate at a fixed dilution.¹⁹ Analytical parameters such as intra-assay and inter-assay reproducibility, and technical and clinical correlation (with the Hemolytic assay) are described by Svar Life Science. Additional information about the design and performance of the test can be found in Seelen et al.¹⁹ and Roos and Wieslander.²⁰ The healthy volunteer reference ranges used by the health center for participant B are those published by Seelen et al.¹⁹: CH50 (≥69), AH50 (30–113), and MBL (0–125). The healthy volunteer reference ranges used by the health center for participant C are CH50 (≥76), AH50 (≥31), and MBL (41–95).

Decreased levels of complement function typically indicate a deficiency or an ongoing immunologic process, leading to increased breakdown of components and depression of complement levels. Increased complement levels are usually a nonspecific expression of an acute phase response.²¹

Graphs were plotted with the healthy volunteer reference intervals for Participant B. Samples from the second DTX401 grade 3 IRR participant (participant B) were collected during the time of the IRR (study day 1) and on study day 150 (∼5 months). A banked, frozen baseline (pre-DTX401) sample was tested after the study day 1 results were available. Correspondence with Ashley Frazer-Able (Exsera BioLabs) regarding a 2018 poster²² confirmed that our sample could be tested. The participant B baseline sample had not undergone any freeze/thaws before testing.

Sex as a biological variable

Sex was not considered as a biological variable in this work. Of the three participants who were further analyzed after they experienced IRRs, two were male and one was female.

Results

DTX401 IRR clinical summary

DTX401 was first assessed in an open-label phase 1/2 study in 12 adults with GSDIa; none of the participants experienced an IRR.¹⁷ In the subsequent phase 3 study,¹⁶ 40 participants received DTX401, and 100% of participants who received a full DTX401 dose reduced their daily cornstarch consumption. Four IRRs occurred. Two were non-serious (grades 1 and 2), and two were serious adverse events (SAEs) (grade 3). The grade 1 and 2 events occurred in participants receiving placebo, and both participants successfully completed their infusions. The SAEs occurred in participants that received DTX401. Both met the anaphylaxis criteria described by Sampson et al.¹⁴ and were reported by the investigators as anaphylaxis. The infusions were not completed.

The first grade 3 SAE IRR was in a 10-year-old male (participant A) who weighed 55 kg and received DTX401 at 500 mL/h (9.1 mL/kg/h, 250 mL over 30 min). The calculated infusion rate was 20.00 × 10¹² vg/kg/h. This participant did not have any infections, illnesses, or disease flares in the month before dosing. He had received a dose of the PEG-containing Pfizer COVID-19 vaccine approximately one year prior to the IRR and tested positive for COVID-19 two months before the event. One minute into the infusion (after administration of ∼6 mL of DTX401), he experienced chest pain, vomiting, tachycardia, and an urticarial rash on his cheeks, arms (more prevalent on the arm where investigational product was being administered), and back. He remained normotensive. The infusion was stopped immediately, and he received ondansetron, acetaminophen, diphenhydramine, and famotidine. A tryptase sample was collected approximately two hours after symptom onset. The local pediatric intensive care unit diagnosed the IRR as anaphylaxis. Steroids and epinephrine were discussed but not administered. The participant was sent to the emergency room for observation. His rash subsided over one hour, and the tachycardia normalized. All symptoms resolved within two hours, and the participant was discharged home well (Table 2).

Table 2.

Summary of grade 3 SAE infusion-related reactions following DTX401 treatment

Participant	Pre- treatment	AE observed (grade)	Infusion rate (vg/kg/h)	Time into infusion (estimated volume)	Infusion re-started?	Age (yr)	Sex/Race	Weight (kg)
A	none	chest pain, vomiting, tachycardia, urticarial rash on cheeks, arms, and back. (grade 3)	20.00 × 1012	1 min (∼6 mL)	no	10	M/ White	55.1
B	acetaminophen and cetirizine	coughing, difficulty breathing (grade 3)	6.67 × 1012	6 min (∼14 mL)	no	12	M/ White	54.1

The second grade 3 SAE IRR was in a 12-year-old male (participant B), who weighed 54 kg and was pretreated with acetaminophen and cetirizine 30 min prior to infusion. The infusion was started at a rate of 167 mL/h (3.1 mL/kg/h, 250 mL over 90 min), with a calculated infusion rate of 6.67 × 10¹² vg/kg/h. Six minutes into the infusion (after administration of ∼14 mL DTX401), he began coughing and had difficulty breathing. He did not have wheezing. His oxygen saturation dropped to 80%, and the infusion was stopped immediately. Oxygen was delivered via a nasal cannula, and inhaled albuterol/salbutamol (Ventolin) and intravenous (IV) diphenhydramine (Benadryl) were administered. He became hypertensive with blood pressure increasing from 115/71 pre-infusion to 159/102 mmHg. An erythematous, reticular rash developed on his neck, abdomen, and legs. Hydrocortisone 100mg IV was given. Tryptase and complement samples were collected 30 min after the initiation of the infusion. Epinephrine was discussed by the clinical team but was not administered. The participant had increased dyspnea and facial flush 30 min after the onset of the initial symptoms (36 min after the start of the infusion), but the participant had good air entry with no wheezing and oxygen saturation of 98% on supplemental oxygen at 6 L/min via nasal cannula. Over the subsequent 30 min, both the dsypnea and rash improved, and supplemental oxygen was discontinued. The rash resolved over a 2-h period. The infusion was not restarted (Table 2). All symptoms resolved within 3 h. The patient was monitored for 24 h and discharged home.

The DTX401 CL301 study protocol was amended to add premedications approximately 30 min before the infusion including acetaminophen, an H1 blocker, and an H2 blocker and to administer the infusion using a slow, staged schedule (Table 1).

DTX401 product quality assessment

An unblinded team was tasked with evaluating any associations between the IRRs and the following critical quality attributes: subvisible particulates, % aggregation, % poloxamer, % empty capsid, % intermediate capsid, HCP, endotoxin levels, in vivo mouse potency, and total protein (e.g., particle titer). The team evaluated both DTX401 and the saline lots that were used as placebo. There was no evidence to suggest that the IRRs were caused by a quality deficiency. Note that the placebo used in this trial did not contain poloxamer 188.

The manufacturing batch records, certificates of analysis, dose homogeneity, change controls, and deviations of the drug substance, drug product, and saline bags administered to the participants impacted by the IRRs were reviewed for a potential cause or relationship with these IRRs, and no events or issues were identified that could have been contributory. The maintenance of storage conditions of these drug substance and drug product batches from manufacturing until dose preparation were reviewed and found to be compliant. With respect to cross-contamination at the Contract Manufacturing Organization for drug substance and drug product, systems and processes to prevent sources of cross-contamination were reviewed and found to be adequate.

An evaluation of the dose preparation and infusion process at the clinics where the IRRs occurred was performed, with special focus on potential endotoxin contamination. It was found that there was compliance with the pharmacy manual, and the personnel were trained adequately. In addition, facilities, equipment, and materials used were adequate. The pharmacy manual process for dose preparation and the infusion method was also reviewed, and there were no findings regarding any potential sources of endotoxin contamination.

DTX401 biomarker studies of participants with grade 3 SAE IRRs

Initial biomarker tests were performed during the IRR events that included tryptase (for participants A and B) and a complement panel (for participant B). After resolution of these events, poloxamer 188 SPT was performed for both participants. Participant B provided an additional sample for complement panel testing on study day 150, and a baseline research sample was also tested. Samples for anti-AAV8 antibodies were tested per protocol. In addition, post-hoc research testing of samples for antibody reactivity to PEG and cross-reactivity with poloxamer 188 was performed.

Tryptase

For participant A, tryptase was measured approximately two hours after symptom onset. The level was 5.9 μg/L, within the reference range of 2.2–13.2 μg/L (Table 3). For participant B, tryptase was measured twice, at 30 min after the onset of symptoms and at 4 h. The levels were 6.8 and 5.3 μg/L, respectively, both within the local reference range of 0–13.5 μg/L (Table 3).

Table 3.

Summary of biomarker testing from two participants that received DTX401 and had infusion-related reactions

Participant	Screening and day 0 anti-AAV TAb result	Tryptase	Skin prick testing to poloxamer 188	Baseline anti-PEG titers	Cross-reactivity to poloxamer 188?	Anti-PEG IgE levels specific IgE (sIgE) concentration (kUA/L)1
A	negative	normal after ∼2 h (5.9 μg/L, normal range 2.2–13.2 μg/L)	negative	not detectable	not applicable	not applicable
B	negative	normal after 30 min (6.8 μg/L) and 4 h (5.3 μg/L), normal range 0–13.5 μg/L	negative	1:12,800	yes	<0.1 (negative)

¹Limit of quantitation 0.1 kU_A/L.

Poloxamer 188 SPT

Poloxamer 188 SPT was performed on participants A and B at three and six months after their reactions, respectively. SPTs to the various dose levels of poloxamer 188 were all negative (Table 3). The positive and negative controls performed as expected.

Anti-AAV8 capsid antibodies (pre-DTX401)

Participants A and B had negative anti-AAV8 capsid antibodies results on the screening and day 0 (pre-DTX401) visits (Table 3).

Anti-PEG antibodies with cross-reactivity to poloxamer 188 and anti-PEG IgE

An exploratory, research-grade method was used to test samples from 43 patients at baseline (pre-DTX401 or pre-placebo) for total anti-PEG antibodies that cross-reacted with poloxamer 188. Ten were positive with titers ranging from 1:100 to 1:12,800 and some had cross-reactivity to poloxamer 188. Participant A did not have detectable anti-PEG antibodies. Participant B had anti-PEG antibodies with a titer of 1:12,800 and these antibodies cross-reacted with poloxamer 188 (Table 3). Samples from all participants with anti-PEG antibodies that cross reacted with poloxamer 188 were tested for anti-PEG IgE. All samples were negative for anti-PEG IgE for Participant B (Table 3).

Complement

Participant B was tested for CH50, AH50, and MBL. CH50 was 110% on day −1 (pre-DTX401), 78% during the IRR on day 1, and 123% on day 150; AH50 was 91% on day −1 (pre-DTX401), 45% during the IRR on day 1, and 104% on day 150 (Figure 1). The levels were approximately 1.5-fold and 2-fold lower on day 1, during the IRR, than on day −1 and day 150. This pattern was indicative of complement consumption during the IRR, despite being within the broad reference range.¹⁹ There was no apparent change in MBL.

Figure 1.

Complement testing for participant B (DTX401) and participant C (UX701)

(A–C) Participant B experienced an IRR after dosing with DTX401. A banked pre-dose sample was tested for CH50, AH50, and MBL complement activity as well as one during the IRR and one 150 days later. (D–F) Participant C experienced an IRR after dosing with UX701. Samples for CH50, AH50, and MBL complement activity were tested at pre-dose (day -4), pre-UX701, approximately 120 minutes after the IRR and just prior to discharge. An AH50 sample was collected for participant C approximately 120 min after the IRR but was mislabeled and thus discarded by the testing lab and a result is not available. No additional samples were collected.

In addition, the day 150 sample from participant B was tested for complement proteins. This sample had elevated C3 and C4 levels (C3 was 1.94 g/L [normal range: 0.79–1.52] and C4 was 0.48 g/L [normal range: 0.17–0.47]).

Anti-AAV8 capsid antibodies (post-DTX401)

Participants A and B developed anti-AAV8 antibodies rapidly and to a high level, despite only receiving a partial dose of DTX401 (Figure 2).

Figure 2.

Anti-AAV8 antibodies in participant A and participant B that received a partial dose of DTX401

The upper limit of the assay is 21,257,640, which is indicated by a dashed horizontal line.

DTX401 infusion metrics

Infusion metrics for DTX401 recipients before and after the implementation of the slow, staged infusion are shown in Table 4. In participants enrolled prior to the implementation of the slow, staged infusion, the mean infusion rate in the first hour was 11.58 × 10¹² vg/kg/h. After the implementation of the slow, staged infusion, the mean infusion rate in the first hour was 0.242 × 10¹² vg/kg/h after. No IRRs were observed after the implementation of the slow, staged infusion. As previously noted, the mean (SD) infusion rate was estimated to be 20.00 × 10¹² vg/kg/h for participant A and 6.67 × 10¹² vg/kg/h for participant B. These two participants were not included in the overall calculations of infusion rates shown in Table 4.

Table 4.

DTX401 infusion metrics

–	Prior to the staged infusion					Staged infusion
	n	mean	SD	min	max	n	mean	SD	min	max
Total dose administered (vg/kg) x 1012	21	9.84	0.93	8.30	13.08	17	9.96	0.86	7.98	11.65
Total infusion duration (min)	21	56.6	19.8	32	104	17	182.6	26.0	96	223
Infusion rate in first hour (vg/kg/h) × 1012	21	11.58	3.86	5.53	18.05	17	0.242	0.017	0.22	0.28

UX701 IRR case—Clinical observations and completed dosing

The IRR in the open-label portion of the Cyprus²⁺ trial was reported in a 20-year-old female (participant C). The management of this event was informed by the IRR experience in the DTX401 study, the last of which had occurred 6 months prior to this event.

The participant received pre-medication with prednisolone 60 mg, acetaminophen 500 mg, cetirizine 10 mg, and famotidine 20 mg approximately one hour prior to dosing. Prior to priming the IV tubing with study drug, the infusion was started at 6 mL/h using the updated infusion protocol that begins at a very slow initial infusion rate. Ten minutes into the infusion, the team then noted that the tubing had not been primed, and so the infusion pump was set to a rate of 999 mL/h to prime the IV tubing with UX701 while the line was inadvertently still connected to the participant. This ran over a 3-min period, during which approximately 50 mL was infused at 9.8 mL/kg/h. This corresponded to a rate of approximately 4.0 × 10¹³ vg/kg/h.

The participant developed flushing on the face and extremities, chills, lightheadedness, dizziness, tachycardia up to 130 bpm, shortness of breath, and abdominal pain. No hypotension was noted. The infusion was stopped, and the principal investigator ordered diphenhydramine IV. Symptoms improved quickly after stopping the infusion, with further improvement of symptoms after diphenhydramine administration. The participant continued to have minor symptoms of tachycardia, slight chills, and abdominal pain. Symptoms were completely resolved approximately 2 h later.

The sponsor and investigator discussed the situation and agreed that the infusion could be restarted at the intended initial infusion rate of 6 mL/h, with close monitoring and a resuscitation cart at the bedside. The infusion was restarted and subsequently advanced according to the staged infusion schedule (Table 1). The infusion was completed uneventfully, and the participant was discharged six hours after completing the infusion.

Laboratory assessments for complement activity were drawn at a screening visit 5 days before the infusion, approximately 2.5 h before the infusion started, and approximately 2 h after the infusion was paused due to the IRR. The final complement samples were collected after completion of the infusion, approximately 5.5 h after the IRR. Each of these time points is shown as a vertical dotted or dashed line in Figure 1.

UX701 biomarker studies

Complement

Testing at approximately 5 h post-IRR demonstrated lower AH50 versus day −1 (pre-GT) (Figures 1D–1F). The AH50 levels were slightly lower than the samples taken 5 days prior and 2 h prior to the infusion. The sample taken approximately 2 h after the IRR was not tested due to error. There were no apparent changes in CH50. Both AH50 and CH50 results were within the assay reference range.¹⁹ MBL activity was elevated in participant C prior to and after UX701 administration. It is unclear why MBL activity would be elevated prior to UX701 administration, although it can be increased in cases of chronic infection or inflammation.²³

Conclusions

Study summary

Two serious IRRs were observed in the GlucoGene study of DTX401 (participants A and B), and one non-serious IRR was observed in the Cyprus²⁺ study of UX701 (participant C). All three events occurred very quickly after initiation of the infusion, with time to onset ranging from one to six minutes. All three events occurred in the setting of a fast infusion rate, ranging from 6.7 × 10¹² to 4 × 10¹³ vg/kg/h.

The results presented herein suggest that the IRRs were rate related, rather than hypersensitivity reactions to active or inactive drug ingredients in DTX401 and UX701 or anti-capsid antibodies. In the case of participant B, anti-PEG/poloxamer 188 antibodies may have contributed to the event (summarized in Table S2), as has been described in vitro.²⁴ Additionally, participant B also had a concurrent diagnosis of atopic dermatitis and elevated complement proteins C3 and C4: this may suggest a possible pre-disposition for complement activation, which has been described previously.²⁵ Upon adoption of the new premedication strategy and a slow, staged infusion, 17 additional participants were treated with DTX401, without IRRs (Table 5). Additional testing for anti-PEG antibodies was performed on these 17 participants, and there was a participant in this group with anti-PEG antibodies, at a titer of 1:1,600, that cross-reacted with poloxamer 188 and also had atopic dermatitis; this participant did not experience an IRR, suggesting improved outcomes with the slowed infusion schedule.

Table 5.

IRR rates

AAV gene therapy	Infusion reactions with fast infusion (30–60 min)	Infusion reactions with slow, staged infusion (∼3 h)
AAV8-G6PC phase 1/2	0/12 (0%)	N/D
AAV8-G6PC phase 3	2/23 (8.7%)	0/17 (0%)
AAV9-ATP7B	1/6 (16.67%)	0/9 (0%)

N/D, not done.

It is notable that while the initial infusion rate that resulted in an IRR for participant B is at the lower end of the range used in the pre-staged group in the DTX401 phase 3 study, it is still substantially more rapid than after the staged infusion protocol was implemented (Table 4). Additional factors, beside infusion rate, that pre-dispose or contribute to IRR sensitivity are not known. The absence of IRRs after implementation of the staged infusion schedule suggests that this approach reduces the IRR risk and therefore allows more patients to potentially receive benefit from a systemically administered AAV gene therapy.

IRRs in AAV gene therapy

IRRs are a common occurrence in AAV gene therapy and several approved IV AAV gene therapies include IRRs, hypersensitivity, or anaphylaxis in the warnings and precautions section of their prescribing information. For Roctavian, an AAV5 gene therapy for hemophilia A, IRRs were observed in 9 patients (7%), including hypersensitivity reactions (4%) and anaphylaxis (1%).¹ Hemgenix, an AAV5 gene therapy for hemophilia B, lists IRRs in Warnings & Precautions (including hypersensitivity and anaphylaxis), and among the most common adverse reactions (33%, n = 19).² Beqvez, an AAVrh74 gene therapy for hemophilia B, lists IRRs under Warnings & Precautions (including hypersensitivity & anaphylaxis).⁵ Zolgensma, an AAV9 gene therapy for spinal muscular atrophy (SMA) in pediatric participants (<2 years old), lists IRRs in the Warnings & Precautions section of the prescribing information.³ Elevidys, an AAVrh74 gene therapy for Duchenne Muscular Dystrophy in individuals at least 4 years of age, also lists IRRs in the Warnings & Precautions section of the prescribing information.⁴

Study limitations

This is a report of three AAV gene therapy recipients who experienced IRRs and is therefore a small study. Although IRRs are a noted safety risk for nearly all approved AAV gene therapies, this case series is the first to provide mechanistic information for IRRs in response to AAVs.

Tryptase

Tryptase is the most abundant protein component of human mast cell secretory granules.^26,27,28,29 Serum levels generally reflect both overall mast cell burden and the extent of acute mast cell activation.³⁰ Tryptase is a commonly used, commercially available test to evaluate suspected anaphylaxis; however, it has limitations. In individuals with food-induced anaphylaxis, even when blood samples are optimally timed, tryptase levels may be within normal limits.³¹ The tryptase rise with anaphylaxis appears to correlate best with severity/the presence of hypotension during the anaphylactic event, and not as well with cutaneous, respiratory, or gastrointestinal signs or symptoms.³¹ Of note, hypotension was not observed in any of the IRRs. Serum tryptase should be measured within 30–120 min of reaction onset when acute hypersensitivity reaction is suspected, especially in the setting of a severe reaction.^32,33,34 This tryptase level can then be compared to baseline level (performed on a pre-reaction sample) or a sample drawn at least 24 h after symptoms resolve. An elevated serum tryptase during a suspected immediate hypersensitivity reaction is thought to provide evidence of mast cell activation and support an IgE-mediated allergic reaction mechanism; however, a normal tryptase or absence of elevation does not exclude the possibility of an IgE-mediated reaction. For individuals with elevated tryptase at baseline, the 20% + 2 formula may be used.^12,35 There is little to no information available on tryptase and AAV gene therapy-associated infusions and reactions.

Skin prick testing (SPT)

SPTs are commonly performed by allergists for the diagnosis of IgE-mediated allergic reactions.³⁶ While commercial extracts are widely available for many food and environmental allergens, this is not the case for many drug allergens. There is little evidence to guide allergists about which SPT drug concentrations are effective but non-irritating or optimal SPT protocols for safe but accurate diagnosis. When faced with a patient with possible drug allergy, allergists often look to case report level data or extrapolate protocols proposed for other drugs. Here, several concentrations of Poloxamer 188 were tested, including 10× the concentration used in the infusion (∼0.03%–0.3%), which is similar to those that have elicited reactions to PEG 20,000.³⁷

Complement testing

Complement activation tests, such as activity assays AH50, CH50, and MBL, can also be difficult to interpret. Due to broad reference ranges, a baseline sample is required for interpretation of results.¹⁹ Measuring the terminal complement complex, sC5b-9, may be easier to interpret due to the very narrow reference range (72–244 ng/mL) in plasma.³⁸

Complement discussion

The complement cascade is an ancient immune-surveillance system that is a global mediator of pathogen elimination, cell homeostasis, tissue development, and repair.³⁹ This extracellular inflammatory cascade has four distinct pathways: (1) the classical pathway (antibody:antigen complexes), (2) the MBL pathway, (3) the alternative pathway (pathogen surface structural repeats), and (4) the granzyme K pathway (tissue-resident T cells in chronic inflammatory diseases).⁴⁰ Each pathway leads to the production of a convertase enzyme that activates the complement proteins C3 and C5, leading to the generation of inflammatory peptides/anaphylatoxins (C3a, C4a, and C5a) and soluble C5b-9, the lytic membrane attack complex (MAC),^40,41 also called the terminal complement complex (TCC). The MAC marks pathogens for destruction and release of peptide mediators of inflammation.

The complement cascade is regulated by two competing cycles,⁴² one that promotes further complement activity (factor B and factor D), and one that resolves it (factor H and factor I).

Enhanced complement activity occurs through a positive feedback circuit that involves binding of factor B to C3b, cleavage of factor B by factor D, formation of C3b-Bb-C3b which promotes further conversion of C3 into C3b, leading to the assembly of C5-C9 into the lytic MAC. This is sometimes called the “amplification loop” (Figure 3). Because they are consumed, continuous production is required to keep the amplification loop active, shown as a continuous cycle of increasing ups and downs in Figure S1. Factor B and factor D are produced by cultured human monocytes,⁴³ hepatocytes,⁴⁴ and in keratinocytes in response to pro-inflammatory cytokines.^43,45

Figure 3.

Schematic overview of the complement cascade in AAV gene therapy, including possible activity from excipients, like poloxamer 188

The classical pathway may be activated by antibodies that bind the capsid or poloxamer 188. These antibodies bind C1q and initiate the cascade by activating C4. The alternative pathway may be activated by direct binding of C3 to the capsid surface, leading to the release of C3a, and pathway activation. The amplification and degradation cycles control the extent of the cascade’s activation.

Control of complement activity is accomplished through the degradation cycle, which begins when factor H binds C3b and acts as a cofactor for complement factor I (CFI) to cleave C3b into iC3b (inactive C3b), in concert with the surface-expressed regulator MCP (CD46). This irreversibly prevents reassembly of the C3 convertase and MAC formation. Factor H and factor I remain unchanged during the degradation cycle and are remarkably stable. This is shown as a slow, rising line in Figure S1. Factor H is rapidly produced (within 30 min) in response to interferon α and interferon γ in human monocytes, in vitro. Factor I is expressed by numerous cell types, including monocytes and endothelial cells,^43,46 and is rapidly induced by interleukin 6 in HepG2 human hepatocytes.⁴⁷ Factor I is stable in serum, in vitro, with similar activity in samples held for 16 h and 63 h at 37C, and minor increases (15%–25%) in factor I levels in whole serum markedly inhibit complement activation by inulin, aggregated IgG, and by low concentrations of cobra venom factor (CVF).^47,48

Therefore, efforts to promote the anti-inflammatory, complement degradation cycle can help control the pathway.

Theoretical mechanistic underpinnings of the slow, staged infusion schedule

Complement activation immediately following AAV gene therapy may be driven directly by the capsid, by anti-capsid antibodies, or by antibodies that bind to excipients, such as poloxamer 188 (Figure 3). In theory, any of these pathways could lead to the clinical manifestations of an IRR.

We hypothesize that upregulation of the steady and stable factor I and factor H during a slow, staged infusion helps promote the anti-inflammatory degradation cycle (Figures 3 and S1). This is likely through pro-inflammatory cytokines produced by monocytes and dendritic cells.⁴⁹ We hypothesize that beginning an infusion with a high AAV capsid concentration does not allow enough time to upregulate sufficient concentration of factor H and factor I, and the complement amplification cycle may become exacerbated, leading to the clinical manifestations of an IRR. However, factors H and I have typically not been studied during infusions, and this theory is untested. Additional work is needed to understand this important pathway.

IRR-related recommendations

We recommend that AAV gene therapy be administered by a slow, staged infusion (Table 1).¹⁵ If a suspected IRR occurs during delivery of an AAV gene therapy at a high infusion rate, pause the infusion; administer supportive care, as clinically indicated, to manage the reaction, and restart at half the prior infusion rate after the IRR has resolved.

Given that this is a preliminary observation, it is important this phenomenon is studied further.

We recommend that baseline tryptase and complement biomarkers be collected in AAV gene therapy recipients. In the case of an IRR, we recommend that both complement and tryptase biomarkers be collected (Table S1). Although a rapid complement test is generally not available to AAV gene therapy centers, this would be a major advancement for monitoring immune biomarkers of AAV safety.

Acknowledgments

The authors wish to thank Brandon Lodigiani for his contributions to this work when he was an employee of Ultragenyx Pharmaceutical Inc, and Binodh DeSilva, Mirna Koscec, and Steve Keve for scientific advice. The authors also thank Jamie Ziobro of Ultragenyx for assistance in the preparation of this manuscript. The Elicit AI agent was used for advanced literature searches in preparation of this manuscript. All primary source data were reviewed for accuracy and referenced herein. This work was funded by Ultragenyx Pharmaceutical Inc.

Author contributions

Conceptualization, B.A.S., E.D.K., and A.A.G.; data interpretation, B.A.S., R. Collis, R. Cahayag, H.M.L., M.G.B., E.D.K., S.R., D.B., M.S.B., J.J.M., and A.A.G.; manuscript writing, B.A.S., R. Collis, R. Cahayag, H.M.L., M.G.B., E.D.K., S.R., D.B., M.S.B., J.J.M., and A.A.G.; manuscript editing, B.A.S., R. Collis, R. Cahayag, H.M.L., M.G.B., E.D.K., S.R., D.B., M.S.B., J.J.M., and A.A.G.

Declaration of interests

B.A.S., R. Collis, R. Cahayag, H.M.L., E.D.K., S.R., and A.A.G. are employees and shareholders of Ultragenyx Pharmaceutical Inc., the sponsor of the clinical trials described herein. M.G.B., D.B., M.S.B., and J.J.M. were investigators in the clinical trials described herein.