The complex, confusing and poorly understood immune responses to AAV‐mediated gene transfer in haemophilia—Is more or less immunosuppression required?

Attempts to achieve a functional cure or amelioration of the severe X linked bleeding disorders haemophilia A (factor VIII deficiency) and haemophilia B (factor IX deficiency) using AAV‐based vectors have been frustrated by immune responses that limit efficacy and durability. The immune responses include adaptive and innate pathways as well as cytokine mediated inflammation, especially of the target organ cells—hepatocytes. Immune suppression has only been partly effective in clinical trials at ameliorating the immune response and the lack of good animal models has delayed progress in identifying mechanisms and developing more effective approaches to controlling these effects of AAV gene transfer. Here we discuss the arguments for and against more potent immunosuppression to improve factor expression after AAV‐mediated gene therapy.


| INTRODUC TI ON
Gene therapy for people with haemophilia has the potential to transform the lives of those who are affected and adeno-associated virus (AAV) has emerged as the leading vector allowing novel genes to be introduced into human cells.However, two major unresolved problems have emerged in AAV gene transfer for treatment of people with haemophilia-the first is the highly variable response in expression levels achieved between individuals.All clinical trials, both preclinical and clinical, show this with response varying over a very wide range to a given dose of vector (expressed as genome per kilogramme body weight of subject).For example, in the pivotal trial of valoctocogene roxaparvovec for people with haemophilia A 1 the median change in factor levels was 22.9 IU per deciliter with an interquartile range of 10.9-61.3,highlighting the variability in response.Efforts to understand and control for this variability are progressing but appear poorly correlated with immune responses, except in as far as pre-existing neutralising antibodies to vector capsid can reduce or prevent vector entry into target cells in vitro and in vivo.The second major problem-the topic of this review-is the variable immune response to AAV vector infusion, leading to an early, mild hepatitis characterised by a modest increase in serum aminotransferases and sometimes associated with a reduction in expression of the novel protein.A gradual decline in expression of the newly introduced factor over the years following therapy is typically seen in patients with haemophilia who have been successfully treated with AAV gene therapy and it is unclear whether this is related to the initial immune response or whether different mechanisms are involved.
Nor is it known whether ameliorating the early immune response will improve the longevity of the treatment.It has proved difficult to understand and manage this problem, which was not observed in preclinical animal models-mice, dogs and non-human primates and progress relies on clinical trial results and inferences from relatively small numbers of subjects.This review will first consider the clinical trial results, which mainly addressed adaptive responses and results of different immune modulating regimes and then summarise the arguments for increasing or reducing the use of immunosuppressive agents.

| Insights into the immune responses to AAV from clinical trials
The first clinical trial to show efficacy in terms of a clinically significant elevation of a clotting factor level after gene therapy was reported in 2006 2 when the vector was administered by the hepatic artery.A subject with severe haemophilia B treated at a high dose with AAV2 serotyped vector containing a factor IX expression cassette achieved a peak level of 12%, in the mild haemophilia range.This level of expression started to fall concomitantly with a rise in alanine aminotransferase (ALT), indicative of liver inflammation.
The level of factor IX fell to baseline by week 12. Subsequent study showed this to have been associated with presentation of AAV capsid-derived peptide by HLA class 1 on the surface of transfected hepatocytes.Adaptive immune response led to CD8 T cell proliferation which then attached to the transfected hepatocytes still presenting vector capsid peptide.This was an early adaptive response to the virus-vector capsid is believed to be present only transiently since the therapeutic DNA sequences do not contain capsid sequences and hence the presented capsid arises from the initial viral inoculum.To-date no studies examining the persistence of the capsid protein have been performed preventing an evaluation of the likely duration of this response.
Building on this result the first successful gene transfer study in people with haemophilia B used oral prednisolone to limit the inflammatory response as soon as any hepatitis (elevation of ALT) was detected. 3The study used a peripheral infusion of AAV serotype 8 at three doses in a total of six subjects and reported that for most cases treated with the higher dose of vector a short course of oral prednisolone curtailed the ALT elevation and improved factor levels.
The authors detected cytotoxic T cells directed against viral proteins in some patients (using ELISPOT assays) and in one subject a strong response coincided with the hepatitis.In a further report of 10 patients with haemophilia B (including the original six subjects with an additional four patients receiving high doses of the vector) similar haematological and hepatological responses were reported with an improvement in circulating clotting factors and an increase in serum ALT in four of six patients receiving high doses of the vector.The hepatitis in all of these patients was treated with prednisolone, using the doses and schedules typically deployed in the management of autoimmune hepatitis (an initial dose of 60 mg per day with a reduction in dose based on the serum ALT concentration).It is instructive to examine the hepatitis in these subjects which were summarised in a follow-up paper 4 and to review the individual responses.This review suggested considerable variability in response to prednisolone therapy with only a few subjects showing an unambiguous response to immunosuppression.From the patients treated with higher doses of vector (2 × 10 12 vector genomes/kg) four received oral prednisolone to control elevation of ALT but two subjects were not treated and had the best responses to therapy.In those who did receive prednisolone all four patients showed a reduction in serum transaminases but only two showed an improvement in factor levels.
The sustained, clinically effective response to gene therapy shown in this study stimulated a rapid expansion of trials using AAV-mediated gene therapy for haemophilia B and subsequently haemophilia A. The haemophilia A clinical trials 1 showed that therapy has a prolonged effect but, as for patients with haemophilia B, transient hepatitis was the commonest side effect of therapy.
Treatment with oral prednisolone appeared to reduce the severity of the liver inflammation and was associated with sustained expression of Factor VIII.However, the proportion of patients developing hepatitis in trials in haemophilia A was increased compared to therapy in patients with haemophilia B and long-term expression of clotting factor differs between the two.For haemophilia A clinical trials show that most patients developed transaminases (>85%) and factor levels appear to decline with time-the 2 year outcomes trial 5 commented 'from week 76 onward, the trajectory of the transgenederived factor VIII activity showed first-order elimination kinetics; the model-estimated typical half-life of the transgene-derived factor VIII production system was 123 weeks (95% confidence interval, 84 to 232)'.For haemophilia B where early liver inflammation is less common (16.9% 6 ) factor production appears to be more durable.It remains unclear whether this is related to the different vectors used, different gene constructs or site of expression of the protein-for haemophilia B the protein is produced in its 'normal' cell (the hepatocyte) whereas for haemophilia A the introduced protein is generated in hepatocytes rather than the usual site of production, liver sinusoidal cells.
The clinical trial data has thus shown that effective gene therapy for patients with haemophilia is possible with AAV vectors, but the associated transaminitis may be related to a reduction in factor expression, and this may be restored, in some patients, by the introduction of prednisolone therapy.These data have led to the licensing of AAV-mediated gene therapies and the therapeutic license includes a recommendation that steroid therapy be introduced to treat the vector induced hepatitis.It is important to note that the use of prednisolone therapy to manage AAV-associated hepatitis was based on a very small number of subjects with haemophilia B and was then incorporated into all future clinical trials without any controlled studies to assess the value and requirement for immunosuppression.Not all patients are able to tolerate prednisolone and, in some subjects included in the pivotal clinical studies, the associated hepatitis did not respond to steroid therapy.Investigators have followed protocols normally used to treat patients with autoimmune hepatitis and therapy with the side effect sparing steroid, budesonide, has been used in a few patients.Alternative regimes include the addition of azathioprine or mycophenolate to the prednisolone or the use of tacrolimus.In the pivotal study in patients with haemophilia A (GENEr8-1) 1 115 of 134 subjects developed a hepatitis and alternatives to prednisolone were used in 39 subjects (29.1%).Tacrolimus was used in 24 patients and mycophenolate was deployed in 13 subjects.The heterogeneity of the regimens used and the small number of patients treated with a wide range of different responses prevents firm conclusions being drawn but the current thinking, based on these limited trial data, is to treat all episodes of AAV-associated hepatitis with immunosuppressive drugs using prednisolone as first-line therapy with the addition or substitution of alternative agents as deemed clinically appropriate.As outlined above the treatment regimes used are based on the assumption that the hepatitis and associated loss of factor expression are caused by a cytotoxic T cell response but the evidence supporting this is relatively weak.
It is important to note that none of the patients in the clinical trials developed a severe hepatitis leading to functional liver damage.All patients with an AAV-associated hepatitis in the trials in people with haemophilia showed a mild degree of liver inflammation without any changes in synthetic function.The purpose of immunosuppression in such patients is to protect the transduced hepatocyte and preserve functional expression of the novel gene.
It therefore seems reasonable to consider withdrawal of immunosuppression if there is no improvement in clotting factor levels, although this approach has not been subjected to any prospective evaluation.However, the clinical trials in patients with haemophilia A have shown that a proportion of patients develop a lowlevel hepatitis in the years following treatment and the long-term impact of this inflammation has not yet been assessed.Prolonged hepatological follow-up of such patients will be required to identify any long-term impact.
An important study evaluating liver biopsies from a small number of patients with haemophilia A who had received valoctocogene roxaparvovec in clinical trials was reported in 2022. 7The study examined liver tissue in patients treated several years earlier and showed, as expected, no evidence of significant liver damage.However, surprisingly, detailed analysis of the liver tissue showed that in some hepatocytes loss of expression of Factor VIII was associated with persistence of the transgene.Until then it was believed that reductions in the expression of AAV-transferred genes were due to loss of the transduced cells-that is, transduced cells were destroyed, perhaps by immune mediated mechanisms.The data showing persistence of nontranscribed DNA indicates that there are additional mechanisms leading to loss of factor expression and this leads to speculation that the initial mild hepatitis seen in the early phases may not be the only, or indeed the main, mechanism underlying short and long-term reductions in efficacy.This paper highlights both our inadequate understanding of the mechanisms involved in cellular control of virally introduced DNA and shows the value of studies based on human biopsies.It is not yet clear what can be done to restore factor levels in cells containing nontranscribed AAV-transferred DNA and future work in this area may identify novel approaches to maintaining factor expression that are less reliant on early control of the immune response.
The clinical trial data to-date thus supports the current regulatory approval and shows that AAV mediated gene therapy combined with an autoimmune hepatitis style treatment of any associated rise in ALT allows long-term expression of the introduced gene.However, in the absence of any randomised trials that do not include immunosuppression it is unclear whether the immunosuppression plays a critical role in maintaining factor levels.Given that a major cause of side effects in the clinical studies was related to steroid therapy there is a strong argument for reducing, or perhaps even abolishing, the use of steroids.On the other hand gene therapy is currently a nonrepeatable procedure (the development of antibodies against the AAV vector precludes retreatment) and therefore any early loss of factor expression due to immune mediated hepatocyte damage cannot be reversed by retreatment and few patients, or their clinicians, would support a programme that exposed them to the risk of treatment failure.In this context the key questions to be addressed are whether it is appropriate to reduce the level of immunosuppression or to increase it and it is unlikely that any trial that did not involve some degree of immunosuppression would be acceptable to the haemophilia community.

| THE C A S E FOR REDUCED IMMUNOSUPPRE SS ION
The use of prednisolone treatment after AAV gene therapy is based on the assumption that the associated hepatitis and loss of factor expression is mediated by an acquired cellular immune response (probably cytotoxic T cell) that is reduced by steroids.The evidence for this is modest and the association between AAV associated hepatitis, factor loss and cytotoxic T cell activity (measured by ELISPOT assay) is relatively weak. 8However, ELISPOT assays are technically difficult, performed on circulating T cells and may not identify transient, low level, intrahepatic T cells that mediate the immune response.
Alternative steroid sensitive immune responses may also be involved in factor loss and the absence of a clear association between in vitro assays and clinical responses does not prove that steroid sensitive mechanisms are not involved.
A number of alternative mechanisms for hepatitis and reduced factor expression have been identified and these are not necessarily steroid sensitive.Innate immune responses to intracellular DNA are mediated by unmethylated dinucleotide CpG motifs in DNA that may function as pathogen associated molecular patterns that activate TLR9 receptors. 9Activation of TLR9 may induce intracellular responses, as well as initiating T cell responses, potentially leading to both hepatitis and factor loss.Although such sequences are minimised in the development of gene therapy inserts it is likely that some intracellular innate responses are activated to a greater or lesser extent and may be involved in short and long-term reduction in factor expression. 10 Over expression of foreign proteins or misfolded native proteins may lead to a cellular stress response leading to cellular death, by apoptosis and in vitro studies confirm that over expression of modified clotting factor VIII in cells does lead to endoplasmic reticulum stress and, ultimately, cell death. 11Hence there are plausible alternative explanations for the observed responses to AAV gene therapeutics that are not necessarily steroid sensitive and may not involve immune activation.
High dose prednisolone, even for a short period of time, is associated with significant side effects and these were the major cause of adverse events in the clinical trials to-date.As noted above there is no evidence of significant liver damage associated with the mild transaminitis seen in patients receiving AAV gene therapy for haemophilia and it is unlikely that failure to introduce immunosuppression will lead to liver damage.Given that, to-date, there is no convincing evidence that larger and more prolonged use of steroids or augmented immunosuppressive therapy increases factor expression it is reasonable to postulate that lower doses of immunosuppression may be associated with fewer side effects without adverse effects on factor level.The Fong study, 7 suggesting that loss of factor is due to intracellular events that may be reversible, introduces the possibility that early loss of factor associated with a mild hepatitis could be overcome by alternative approaches that restore transcription of virally introduced DNA.Hence a case can be made for reducing the level of immunosuppression and improving factor expression by less toxic drugs.However, the case for reducing immunosuppression is purely theoretical and may be dependent upon novel, as yet unexplored, 'rescue' therapies that restore expression from silenced AAV introduced DNA.It is the author's view that reduced immunosuppression approaches should not be introduced outside the context of a controlled clinical trial.

| THE C A S E FOR ENHAN CED IMMUNOSUPPRE SS ION
Although a proportion of patients with haemophilia who are treated with current AAV gene therapies derive sustained long-term benefits circulating levels of clotting factor diminish with time (particularly in patients with haemophilia A) and it seems likely that many patients will require reintroduction of alternative treatments after several years.It is reasonable to assume that increasing the early efficacy of AAV-mediated gene therapy will increase the longevity of the therapy and if some of the decline in efficacy is related to immune-mediated damage to transduced hepatocytes then early, more aggressive immunotherapy may improve both the short-and long-term clinical outcomes.Any initial increase in efficacy must, however, be limited as there is a risk of excess clotting if the initial increases in clotting factors are too great.There is thus a theoretical argument for enhanced immunosuppression to improve the effectiveness of gene therapy.
Current treatment guidelines recommend introducing prednisolone therapy only after a hepatitis has developed.It is reasonable to question whether treating everyone with prophylactic steroids will prevent any liver damage and lead to improved factor expression.A clinical trial examining this proposal is under way and the results were recently presented at an international conference-(Ozelo EAHAD, Frankfurt 2024) and show that prophylactic prednisolone may be detrimental.The full publication of these important data will increase our understanding of the optimal use of steroids in gene therapy and we look forward to the release of further details of this study.
A number of animal models have recently been developed that support the use of more aggressive immunosuppression.
To be relevant to clinical gene therapy the animal model should share key features of the clinical trials with loss of expression of the transgene not simply due to a foreign protein elicited adaptive immune response, and such models have recently been presented.Herzog and colleagues developed a haemophilia A mouse model having the key features noted above 12 and in this model they showed that a combination of IL_15 inhibition and rapamycin attenuated the time-dependent reduction in clotting factor levels seen in control, untreated animals.An important observation was that this approach seemed to increase the proportion of transcriptionally active AAV gene inserts suggesting that this may reduce the 'gene silencing' that has been reported in humans.
However, murine immune responses may not mirror human responses and the highly selected mouse strain used in these experiments had an atypical, enhanced antibody response to the introduced Factor VIII, making it difficult to determine whether these results can be applied more widely.Readers interested in further details of our current understanding of the complex interactions between the host and virus in the context of gene therapy for haemophilia can find further details in the recent review by Pierce and colleagues. 13

| CON CLUS IONS
The major side effect of AAV-mediated gene therapy in people with haemophilia is the associated hepatitis.Although the liver injury is usually mild and clinically insignificant it may be associated with loss of factor expression.Current strategies are based on protocols developed in the early clinical trials with very small numbers of patients and show that the response is heterogenous and that the pathophysiology is by no means clear.Studies based on liver histology in patients who have undergone gene therapy have already provided novel insights into the mechanisms underlying long-term reductions in factor expression and it is to be hoped that future studies will include an evaluation of liver histology and allow a more rational approach to effective control of the AAV gene therapy associated hepatitis.For the present the arguments for augmented or reduced immunosuppression are finely balanced and based on personal opinion rather than evidence.It is the authors view that alternative approaches to immunosuppression are best addressed in the context of controlled clinical trials.