Osteoarthritis Gene Therapy in 2022

Structured Abstract

Purpose of Review:

To assess the present status of gene therapy for osteoarthritis (OA).

Recent Findings:

An expanding list of cDNAs show therapeutic activity when introduced into the joints of animals with experimental models of OA. In vivo delivery with adenovirus or adeno-associated virus is most commonly used for this purpose. The list of encoded products includes cytokines, cytokine antagonists, enzymes, enzyme inhibitors, growth factors and non-coding RNA. Elements of CRISPR-Cas have also been delivered to mouse knees to ablate key genes. Several human trials have been initiated, using transgenes encoding transforming growth factor-β, interleukin-1 receptor antagonist, interferon-β, the NKX3.2 transcription factor or variant interleukin-10. The first of these, using ex vivo delivery with allogeneic chondrocytes, gained approval in Korea which was subsequently retracted. However, it is undergoing Phase III clinical trials in the US. The other trials are in Phase I or II. No gene therapy for OA has current marketing approval in any jurisdiction.

Summary:

Extensive pre-clinical data support the use of intra-articular gene therapy for treating OA. Translation is beginning to accelerate and six gene therapeutics are in clinical trials. Importantly, venture capital has begun to flow and at least seven companies are developing products. Significant progress in the future can be expected.

Introduction

Although osteoarthritis (OA) has high heritability, variously estimated between 40% and 60% [1], this reflects the influences of a large number of genes, each of which has a small individual effect. A recent genome wide association meta-analysis, for example, identified 100 different loci associated with OA [2]. Thus, OA is clearly not a good candidate for traditional gene therapy that targets Mendelian disorders, where a single gene defect is responsible for a given disease. Nevertheless OA is a good candidate for gene therapies in which local gene transfer to the affected joint is used as a type of drug delivery system, where the active agent is the product encoded by the transgene [3].

Intra-articular therapy is an attractive strategy for OA because the disease primarily affects a limited number of weight-bearing joints and has no known important extra-articular sequelae. This approach delivers the therapeutic agent directly to the diseased joint, thus enhancing its effectiveness while greatly lowering the risk of adverse events and reducing costs. Local delivery by intra-articular therapy has become particularly important since the development of biologics. The latter, typically proteins, including antibodies, do not enter joints very efficiently following systemic delivery because of the sieving effect of the fenestrated endothelium of the sub-synovial blood vessels [4]. Although intra-articular injection obviates this limitation, there are two additional barriers to delivery: rapid efflux of the injected material via lymphatic drainage and, for drugs acting directly on chondrocytes, the dense extra-cellular matrix (ECM) of the cartilage [4]. No existing technology has succeeded in overcoming these barriers. In 1992 we proposed the novel approach of using gene delivery to target anti-arthritic proteins to joints [5]. The literature prior to 2018 has been comprehensively reviewed in reference 3. The present review summarizes the newer literature with an emphasis on clinical translation and clinical trials.

Gene Delivery to Joints by Intra-articular Injection

As reviewed extensively elsewhere [3], both in vivo and ex vivo strategies have been successful in delivering genes to joints in laboratory animals and, as described below, have been employed in human clinical trials. Reflecting the technology of the time, ex vivo delivery was the first strategy to be developed [6] and was used in the first delivery of a gene to a human joint [7]. However, with one notable exception (see below), this approach has been discontinued for clinical development with the advent of improved vectors allowing efficient in vivo delivery by intra-articular injection with extended periods of local transgene expression. Of the numerous vectors tested pre-clinically as vehicles of in vivo gene delivery to joints, adeno-associated virus (AAV), high-capacity (HC) adenovirus and plasmid DNA are in clinical trials. Self-complementing AAV is the only vector with the confirmed ability to achieve sustained, high levels of intra-articular transgene expression ([8] and Ghivizzani et al, unpublished). In addition, AAV is able to transduce chondrocytes in diseased joints more effectively than other viral vectors, most likely due to its substantially smaller capsid size. Additional important factors that determine the choice of vector are safety, the size of the transgene and its regulatory elements, ease and cost of manufacturing, the immune response to the vector and the ability to re-dose [9].

Preclinical Evaluation in Animal Models of Osteoarthritis

The choice of therapeutic transgene(s) for delivery to joints with OA remains controversial, which is not surprising given the complexity of OA, inexact understanding of its pathophysiology and its different phenotypes. A large number of cDNAs have been tested in animal models (Table 1).

Table 1:

Use of intra-articular gene therapy in animal models of osteoarthritis

Transgene	Vector In/Ex vivo	Model	Result	Reference
IL-1Ra	Retrovirus Ex vivo	Dog ACL transection	Reduced early loss of cartilage	[37]
IL-1Ra	Plasmid In vivo	Rabbit partial meniscectomy	Reduced osteophytes, cartilage loss	[38]
IL-1Ra	AAV In vivo	Horse osteochondral chip	Reduced lameness, cartilage loss, synovitis	[8]
IL-1Ra	Adenovirus	Horse osteochondral chip	Reduced lameness, cartilage loss, synovitis	[36]
IL-1Ra	HC adenovirus In vivo	Mouse Horse osteochondral chip	Mouse: Reduced cartilage loss, osteophytes, hyperalgesia Horse: reduced lameness, cartilage loss, synovitis	[10]
IL-1Ra	HC adenovirus In vivo	Rat ACL transection	Reduced cartilage loss	[11]*
Lubricin	HC adenovirus In vivo	Mouse ACL transection	Reduced cartilage loss	[13]
Lubricin	AAV In vivo	Rabbit ACL transection	Reduced cartilage loss	[14]
Lubricin and IL-1Ra	HC adenovirus In vivo	Mouse DMM, CLT	Stronger anti-OA effect than either gene alone	[15]
IL-1Ra, sTNFR1 alone, combined	Adenovirus In vivo	Rabbit MCL transection, meniscectomy	IL-1Ra, not sTNFR1, reduced cartilage loss. Combination stronger	[16]
TGF-β1	Retrovirus Ex vivo	Rat MIA	Reduced pain, cartilage loss	[17]
Variant IL-10	Plasmid In vivo	Canine Naturally occurring	Reduced pain	[19]**
BMP-4 and sFlt-1	Retrovirus Ex vivo	Rat MIA	Improved cartilage repair	[39]
γ-glutamyl carboxylase	Lentivirus In vivo	Lapine ACL transection	Reduced inflammation, cartilage loss	[40]
Kallistatin	Adenovirus In vivo	Rat ACL transection	Reduced cartilage loss	[41]
Cytostatin-C	AAV In vivo	Rabbit ACL transection	No effect	[42]
Thrombospondin-1	Adenovirus In vivo	Rat ACL transection	Reduced inflammation, cartilage loss	[43]
Hemoxygenase-1	AAV In vivo	STR/ort Mouse Spontaneous	No effect on cartilage loss	[44]
Dkk-1; Wnt10b	AAV In vivo	Rat DMM	Dkk-1 reduced osteophytosis; Wnt10b no effect.	[45]
Ras homolog enriched in brain (RHEB)	Adenovirus In vivo	Mouse DMM	Reduced cartilage loss	[46]
CRISPR/cas vs NGF, IL-1β, MMP13	AAV In vivo	Mouse DMM	Ablating NGF reduced pain; IL-1β, MMP13 preserved structure	[20]
shRNA vs MIP-γ1	Lentivirus In vivo	Mouse ACL transection	Reduced cartilage loss	[47]
miR-93–5p	Lentivirus In vivo	Rat ACL transection	Reduced cartilage loss	[48]
miR-140–5p	Lentivirus Ex vivo	Rat DMM	Improved cartilage repair	[49]

Abbreviations

ACL: Anterior cruciate ligament

BMP: Bone morphogenetic protein

Cas: CRISPR-associated protein

CLT: Cruciate ligament transection

CRISPR: Clustered Regularly Interspaced Short Palindromic Repeats

DMM: Destabilization of the medial meniscus

HC: High capacity

IL Interleukin

IL-1Ra: IL-1 receptor antagonist

MIA: Mono-iodoacetate

MIP: Macrophage inflammatory protein

MMP: Matrix metalloproteinase

NGF: Nerve growth factor

sFlt-1: Soluble fms-like tyrosine kinase-1

shRNA: short hairpin RNA

sTNFR: Soluble tumor necrosis factor-α receptor

TGF-β: Transforming growth factor-β

The interleukin-1 receptor antagonist (IL-1Ra) has been the most studied reflecting the status of IL-1 as a popular, albeit controversial, therapeutic target for OA. The most recent publications in this context report safety and efficacy in mice [10], rats [11*, 12] and horses [8, 10] using AAV and HC-adenovirus vectors. These data support two current phase I clinical trials, described in the next section.

The lubricating glycoprotein lubricin, also known as proteoglycan 4 or PRG4, has shown efficacy when delivered by HC-adenovirus (also known as helper-dependent or gutted adenovirus) as a transgene in mouse models of OA [13]. However, a recent paper using AAV delivery in a rabbit model described loss of efficacy when gene transfer was delayed by 2 weeks after surgery [14]. Stone et al. have reported that a combination of IL-1Ra and PRG4 cDNAs gives a stronger effect than either gene alone [15]. Likewise, a combination of IL-1Ra and soluble tumor necrosis-α receptor-1 (sTNFRI) cDNAs has shown greater efficacy than either alone in a rabbit model [16]. It is interesting that sTNFRI had no effect by itself, but potentiated the action of IL-1Ra.

A recent publication describes efficacy when injecting retrovirally transduced human chondrocytes over-expressing transforming growth factor-β₁ into the knee joints of rats with OA induced by mono-iodoacetate (MIA) [17]. According to this paper, the xenografted human cells reduced pain and preserved cartilage by inducing macrophages in the joint to switch from a M1 to a M2 phenotype. This paper was published after clinical trials had started (see next section), with safety having been demonstrated in mice, rabbits and goats [18]. Later developments have cast doubt on the identity of the cells used in these studies (see next section).

Most pre-clinical development of arthritis gene therapy has focused on the use of viral vectors because transgene expression using non-viral delivery is low and transient, while many non-viral vectors are inflammatory. Nevertheless, Watkins et al [19**] recently reported success in delivering plasmid DNA encoding a long-acting variant of IL-10 to the stifle joints of dogs with naturally occurring OA. The treatment reduced pain and is now in human clinical trials.

Additional transgenes evaluated are listed in Table 1. One novel approach uses AAV to deliver elements of the CRISPR/Cas system to ablate specific genes whose products are involved in the intra-articular pathophysiology of OA [20]. In a mouse model of OA, ablation of nerve growth factor (NGF) reduced pain but enhanced structural deterioration of the joint. Ablating IL-1β or matrix metalloproteinse-13 (MMP-13) preserved structure but was less effective than NGF ablation in reducing pain [20].

Of the 24 pre-clinical studies listed in Table 1, 20 were performed in animal models that required surgical destabilization of the joint. While these models are likely to be relevant to post-traumatic OA, their relevance to idiopathic OA, the majority manifestation, is unclear [21]. Likewise, two of the listed studies used MIA, an inhibitor of glycolysis, to induce OA. Its relevance to human OA is also questionable, although it is frequently used as a pain model. It should also be noted that few of these experiments used large animal models of OA, which may be more reflective of human OA than experimental OA in rodents and rabbits.

Relatively little attention has been paid to regulation of transgene expression. Most studies have used a strong constitutive promoter, such as the cytomegalovirus (CMV) immediate early promoter/enhancer. Elements responsive to the transcription factor nuclear factor kappa B (NF-κB) have been most widely explored in the context of inducible transgene expression.

Most research has focused on intra-articular gene delivery. However, there has been recent interest in systemic delivery as a way of addressing systemic issues that engender OA. In this context AAV-mediated follistatin gene transfer, delivered via intra-venous injection to mice fed a high-fat diet, reduced body weight while also lowering levels of inflammatory mediators and diminishing susceptibility to OA induced by surgical destabilization of the medial meniscus (DMM) [22*]. Translating this interesting approach will run into the issues of manufacturing and cost illustrated by the example of Zolgensma, a systemic, AAV-based, in vivo gene therapy for spinal muscular atrophy that costs $2 million per dose [23]. For a common condition such as OA this may be unsupportable, but if the gene therapy also addresses obesity, metabolic syndrome and related conditions there could be room for debate. However, recent safety concerns when using high doses of AAV systemically will also be a major factor [24].

Clinical Trials in Osteoarthritis Gene Therapy

There have been several clinical trials of OA gene therapy (Table 2). Transgenes evaluated in these trials encode TGF-β₁, interferon (IFN)-β, IL-1Ra, the NKX3.2 transcription factor and a variant of IL-10.

Table 2:

Clinical trials in the gene therapy of osteoarthritis

Transgene	Vector (In vivo/Ex vivo)	Phase	NCT identifier	Status	References
TGF-β1	Retrovirus Ex vivo	I, II, III	02341391 02341378 02072070 01671072 00599248 03291470 03203330	Completed Completed Completed Completed Completed Recruiting Recruiting	[25–27]
IFN-β#	AAV In vivo	I	02727764	Active, not recruiting	[29]**
IL-1Ra	AAV In vivo	I	02790723	Completed	[30]
IL-1Ra	HC-Adenovirus In vivo	I	04119687	Active, not recruiting	[31]
IL-10	Plasmid In vivo	I, II	03477487 04124042	Completed Completed
NKX3.2	AAV In vivo	I	04875754 05454566	Active, recruiting Not yet recruiting

^#This trial targeted inflammatory hand arthritis, which could be caused by OA or RA

The first human trials delivered TGF-β₁ to knee joints with OA in an ex vivo fashion using allogenic chondrocytes. Allogeneic cells obviate the inconvenience and costs associated with the use of culture expanded, autologous cells, but bring significant risk of immune rejection. The chondrocytes were derived from the digit of an infant with polydactyly and were retrovirally transduced to over-express TGF-β₁. Because of the potential for insertional mutagenesis, prior to injection the transduced cells were irradiated at a dose that inhibited cell division, and thus the potential for malignancy, while permitting continued expression of TGF-β. Before injection, the transduced cells were mixed with untransduced chondrocytes from the same donor to sustain and amplify the effect.

One issue with ex vivo gene delivery to joints is the short intra-articular dwell time of the injected cells. Nevertheless, the primary endpoints of Phase I [25], II [26] and III [27] trials were met. These endpoints were mainly patient reported outcomes, but an MRI study suggested reduced disease progression and synovitis, while bone marrow lesions, meniscal damage and osteophytosis were unaffected [28]. In 2017 the South Korean authorities gave marketing approval to the therapy as the drug Invossa. This was the first gene therapeutic approved in Korea and the first for OA anywhere. This approval was revoked in 2019 when it came to light that the genetically modified cells were predominantly HEK293 cells, not chondrocytes. However, two Phase III trials have been allowed in the US (NCT03291470; NCT03203330) on the basis that the earlier US trials, unlike those in Korea, used HEK293 cells from the beginning. These trials are recruiting subjects with OA of the knee (Table 2) while the status of Invossa in Korea is being determined.

Other trials in OA use AAV for the intra-articular delivery of IFN-β, IL-1Ra or NKX3.2. Data from the IFN-β study have recently been published [29**]. AAV serotype 5 expressing IFN-β, under the transcriptional control of a NF-κB responsive promoter, was injected into joints of the hand with inflammatory hand arthritis; this could result from either OA or rheumatoid arthritis (RA). Although the trial was designed to dose 12 subjects, it was stopped after 4 subjects had been injected because of severe tenosynovitis in 3 of the 4 individuals. It is interesting that the 3 subjects whose adverse reactions led to discontinuation had OA, whereas the patient with only a mild reaction had RA. The adverse reactions were delayed, appearing between 4 and 12 days after injection of the vector, and persisted for 5 weeks to 4 months. They may have been related to the high dose of vector that was administered: 1.2 × 10¹² viral genomes (vg) in the carpometacarpal joint and 0.6 × 10¹² vg in the proximal interphalangeal joint. Patients were screened for neutralizing antibodies (Nab) against AAV5 to exclude those with pre-existing high titers. Nab to AAV5, but not IFN-β, were generated after injection of the virus without a detectable T-cell response to the vector. Shedding of virus was limited to transient, small increases in the blood and saliva

Data from the trials using AAV or HC-adenovirus to deliver IL-1Ra have not been published in the refereed literature, although each has been the subject of a recent abstract [30, 31]. The former uses AAV serotype 2.5 and a CMV promoter. Intra-articular injection of this vector produced no severe adverse events. One subject, who had pre-existing Nab to AAV2.5, experienced a mild/moderate knee effusion following injection which resolved with ice and rest [30]. Pre-existing neutralizing Nab to AAV2.5 have been measured previously in the synovial fluids and sera of patients with OA [32], but only 16% of patients had high titer Nab. Titers in synovial fluid correlated with those in sera, indicating that circulating levels of Nab could be measured to predict titers in synovial fluid.

Preliminary data from the first 5 subjects dosed with the HC-adenovirus vector encoding IL-1Ra suggest that the treatment is well tolerated with evidence of a clinical response in some patients [31]. The cassette delivered by the HC-adenovirus has a NF-κB inducible promoter to restrict transgene expression to inflammatory episodes. In this context it is interesting that higher levels of in vivo IL-1Ra expression are also seen with the AAV vector under inflammatory conditions [8], suggesting that the CMV promoter used in this vector also has an inducible component.

Recently two additional trials have been added to clinicaltrials.gov involving intra-articular injection of an AAV vector serotype 5.2 containing cDNA encoding the NKX3.2 transcription factor, which is responsible for maintaining the chondrocytic phenotype by preventing progression to hypertrophy [33]. Although little supporting information appears in the literature, one of these trials is actively recruiting OA patients (Table 2).

Finally, it is encouraging that testing of the IL-10 variant plasmid has progressed to a Phase II trial that is recorded as “completed” on ClinicalTrials.gov (NCT04124042). No outcome data are given, so it is not possible to evaluate the results. This variant of IL-10 does not appear to have been published in the scientific literature, but it may be one described in a US patent application that describes a combination of IL-10 with the IL-10R1 component of its receptor [34].

Perspectives

Gene therapy approaches to treating OA are now attracting serious attention. Several clinical trials have been initiated and one has completed Phase II; Phase III trials have been completed in Korea and are in progress in the US (Table 2). Venture capital is beginning to flow and there are at least seven companies developing gene therapeutics for OA. These advances auger well for accelerating development of the field.

Although considerable progress has been made in overcoming the technological hurdle of achieving long-term transgene expression in joints, there is no consensus on the choice of transgene and promoter. Constructs expressing more than one therapeutic gene product may have greater potency, and it is also worth considering combination therapy in which gene therapy is used in conjunction with other medical or surgical approaches. Because OA is a very heterogeneous disease, different types of OA may need to be matched with specific gene therapeutics, in which case a diagnostic test for patient selection would be of great value. In the case of IL-1Ra, for example, different haplotypes of the IL1RN gene predicting greater severity and quicker progression of disease could identify patients who respond best to IL-1Ra gene therapy [35*].

Key Points.

• Advances in vector technology now permit in vivo gene delivery to joints with extended periods of transgene expression. The development of self-complementing AAV vectors has been particularly valuable in this regard.

• A number of different cDNAs have shown therapeutic effects in animal models of OA using rodents and rabbits, but few have been evaluated in larger animals that may reflect better human joints with OA.

• Six different gene therapeutics for the intra-articular treatment of OA are in clinical trials.

• No gene therapy for OA is presently approved in any jurisdiction, but increasing investment from academia, biotechnology and venture capital groups provides optimism about future progress.