Challenges and strategies toward oncolytic virotherapy for leptomeningeal metastasis

Abstract

Meningeal metastasis (LM) is commonly seen in the advanced stages of cancer patients, often leading to a rapid decline in survival time and quality of life. Currently, there is still a lack of standardized treatments. Oncolytic viruses (OVs) are a class of emerging cancer therapeutics with the advantages of selectively replicating in cancer cells, delivering various eukaryotic transgenes, inducing immunogenic cell death, and promoting anti-tumor immunity. Some studies applying OVs intrathoracically or intraperitoneally for the treatment of malignant pleural and peritoneal effusions have shown promising therapeutic effects. If OVs could be applied to treat LM, it would bring significant survival benefits to patients with LM. In this review, we analyzed past research on the use of viruses to treat meningeal metastasis, summarized the efficacy and safety demonstrated by the research results, and analyzed the feasibility of oncolytic virus therapy for meningeal metastasis. We also summarized the existing data to provide guidance for the development of OVs that can be injected into the cerebrospinal fluid (CSF).

Introduction

Leptomeningeal metastasis (LM) refers to the involvement of the meninges by malignant tumors through hematogenous spread, seeding via cerebrospinal fluid (CSF), or direct invasion by adjacent tumors, and it is a serious complication often seen in the advanced stages of cancer patients [1]. Radiologically, it manifests as enhancement of the pia mater on central nervous system (CNS) Computed Tomography/Magnetic Resonance Imaging (CT/MRI) scans [2]. Patients often suffer from increased intracranial pressure and compression of brain tissue by CSF accumulation, leading to symptoms such as headache, vomiting, intellectual disabilities, and gait disturbances. When the brain parenchyma and meninges are involved, symptoms such as seizures, meningeal irritation signs, and motor disabilities may occur. Additionally, it can affect intracranial nerves causing impaired nerve function such as facial numbness, speech difficulties, etc.; involving the cauda equina can lead to cauda equina syndrome such as incontinence of urine and feces [3]. Virtually all malignant tumors can develop leptomeningeal metastases. Notably, the highest incidence of LM is observed in breast cancer (12–35%), lung cancer (10–26%), and melanoma (5–25%) [4, 5]. The incidence and prevalence of LM may increase due to a combination of factors, including improvements in imaging technology, lower thresholds for initiating diagnostic tests, and improvements in systemic control and prolonged survival of patients with malignant tumors [3].

Without treatment, patients with LM have a life expectancy of approximately 6–8 weeks. For those who undergo active treatment, the median survival ranges from 3 to 11 months [6]. Current treatment modalities include radiotherapy, systemic chemotherapy delivery, small molecule targeted drug therapy, and intracranial injection of chemotherapeutic drugs. Radiotherapy is commonly used to rapidly alleviate symptoms and treat large volume disease, but it has not been definitively proven to improve survival rates [7]. Systemic drug delivery therapy requires the drug to have strong penetration of the blood-CSF barrier. Otherwise, very high systemic doses would be required to achieve therapeutic drug concentrations in the CSF [8]. Intracranial injection chemotherapy helps drugs bypass the blood-CSF barrier (BCSFB); however, drugs administered intrathecally will only spread within a few millimeters of tissue cells adjacent to the CSF, limiting their effectiveness for patients with large volume meningeal metastases [9]. For patients with LM, better treatment methods still need to be developed to improve the quality of life and survival.

Oncolytic virotherapy (OVT) is a promising emerging cancer immunotherapy that leverages the natural ability of certain replicating viruses to infect and preferentially lyse tumor cells while leaving non-tumor cells intact [10]. The viruses are sourced from naturally occurring viruses in the environment and selected based on their innate ability to induce immunogenic cell death (ICD) in cancer cells, with their tumor selectivity enhanced through genetic engineering or specific viral vectors to promote replication, limit pathogenicity, weaken immunogenicity, and reduce neutralizing antiviral immune responses [11]. OVs exert their lytic effects on both primary and metastatic tumor sites through local or systemic administration [12]. With advancements in preclinical and clinical research on OVT, various administration methods for OVs have been explored. Notably, local administration involving the injection of OVs into natural body cavities has shown good therapeutic efficacy and safety in studies treating malignant ascites, ovarian cancer, etc., by locally infiltrating to treat tumors on the cavity walls or metastatic tumors within the cavity [13,14,15]. Therefore, whether OVs can be used to treat leptomeningeal metastases and malignant CSF caused by LM through intracranial injection is a direction worth exploring. More than a decade ago, scientists explored the feasibility of using viral gene therapy and OVT for LM [16]. However, with the advent of chemotherapy drugs and failures in clinical trials, this viral treatment method for LM has gradually been abandoned [17]. Now, with the burgeoning of OVT in the field of tumor immunotherapy, coupled with the exploration of immuno-combination therapies, whether this will propel the application of OVT in the treatment of LM is a question worth paying attention to [18,19,20,21].

In this review, we first summarize the past research achievements in the treatment of LM using viral gene therapy, as well as the experience learned from clinical experience with oncolytic virus treatment of brain tumors. We also describe the latest research progress in current oncolytic virus administration methods. A better understanding of how to optimize the design of oncolytic virus drugs in the treatment of LM will enhance the clinical management of patients with LM and other cancer patients.

Early viral therapy for LM

The use of OVs to treat tumors dates back to 1904 when The Lancet reported a case of a 42-year-old chronic leukemia patient who experienced a reduction in diseased white blood cells and improvement in condition after severe influenza [22]. In 1912, an Italian doctor, Depace, observed a cervical cancer patient whose tumor naturally shrank and regressed after receiving a rabies vaccine [23]. This sparked research into the use of viruses to treat tumors, marking the beginning of the history of viral therapy for cancer.

Subsequently, more incidents of tumor regression due to natural viral infections were reported, prompting researchers to conduct hundreds of studies using viruses to treat tumors between 1950 and 1970 [24]. However, until the mid-1980s, prior to the molecular biology revolution, these clinical trials utilized naturally occurring viruses and differed from current clinical trial standards, making it difficult to assess the true efficacy of the trials. On the contrary, due to the unknown mechanisms of viral pathogenesis, many viruses had fatal consequences in terms of safety. For example, there were high risks of encephalitis from diseases such as West Nile, Uganda, Dengue, and Yellow fever [25].

With the development of molecular biology and an enhanced understanding of viral biology, people began to genetically modify viruses for cancer therapy. In 1991, Martuza R L and colleagues genetically engineered a herpes simplex virus (HSV) with a completely inactivated thymidine kinase (tk) gene. In animal experiments, inoculation of this virus into tumors extended the survival of nude mice with intracranial U87 glioblastoma [26]. This led to the exploration of different viral modification strategies for the treatment of various tumors.

In the 1990s, researchers initiated studies on the use of viral gene therapy for LM, rather than OVT. The early strategy was gene-directed enzyme prodrug therapy (GDEPT), which utilized replication-deficient viruses to transduce foreign genes into cancer cells, increasing the selectivity of cancer cells to drugs, where the virus served merely as a vector capable of delivering therapeutic genes [27]. The treatment consists of two steps: (I) delivering the gene encoding the foreign enzyme to the target cells; (II) administering a prodrug that is converted into a toxic drug by the newly expressed enzyme in the target cells [28]. For example, by using viral transduction to introduce herpes simplex virus thymidine kinase (HSV-tk), followed by treatment with the nucleoside analog ganciclovir(GCV), the drug sensitivity of the tumor is increased then they are killed [29]. In this treatment strategy, only tumor cells that have been transduced with the HSV-tk can convert GCV into a toxic form, while thymidine kinase in normal cells does not have this ability, thus GCV has almost no toxicity to cells that have not been transduced with the HSV-tk [16]. The GDEPT method is similar to chemotherapy; however, the chemical drug targets from within the cancer cells and is generally less toxic to normal cells [30].

Early gene therapy for LM was designed based on this, by injecting viruses carrying HSV-tk into the brain ventricles or intrathecally, followed by intraperitoneal (i.p.) or intravenous (i.v.) administration of the antiviral drug GCV, showing good anti-tumor effects in both in vivo and in vitro experiments. The tested viruses including retroviruses (RVs), adenoviruses(ADs), and HSV.

In 1994, Ram constructed a 9L syngeneic gliosarcoma rat model of leptomeningeal neoplasia, followed by intrathecal injection of retroviral producer cells, achieving in situ transduction of tumor cells by retroviruses carrying HSV-tk in the CSF, and then treating this model with i.p. GCV. Compared with the control group that did not receive vector-producing cells, i.p. GCV administration resulted in a significant prolongation of survival in rats given injections of thymidine kinase vector-producer cells. Injection of producer cells coinfected with the 4070A retrovirus did not improve antitumor efficacy [31].

In 1996, Vincent and colleagues intrathecally injected recombinant adenoviruses carrying HSV-tk into a rat model of LM of 9L cells, followed by i.p. administration of GCV, and the results showed that rats treated with HSV-tk and subsequent GCV had a significantly longer symptom-free latency than all control groups [32, 33].

The same year, Kramm used a herpes simplex virus vector/GCV strategy to treat a rat model of LM of 9L gliosarcoma, and 90% of the animals treated with the combined therapy achieved tumor-free, long-term survival (LTS), while only 30% of the animals treated with the vector alone achieved LTS, and only 10% of untreated animals achieved LTS [34, 35]. From the experimental results, this treatment strategy showed good therapeutic effects in preclinical animal models. The cytotoxic effects of this strategy are mainly twofold: (I) suicide effect, where the newly expressed enzyme in tumor cells converts the prodrug into a toxic drug, inducing cell death. (II) Bystander effect, where in most tumors, the virus can only transduce surface tumor cells near the injection site and cannot transduce cells inside the tumor. However, these tumor cells that have been transduced with the suicide gene generate toxic substances that diffuse to the side and can kill neighboring tumor cells [36,37,38]. This eliminates the need to transduce each tumor cell with the viral gene to achieve tumor regression [16]. (III) The process of cancer cell death initiates a danger signal by releasing damage-associated molecular patterns (DAMPs), which play a role in immature dendritic cells by increasing antigen uptake, maturation, and presentation to cytotoxic T lymphocytes, triggering the activation of anti-tumor immunity [30, 39, 40].

At the beginning of the twenty-first century, researchers began to explore the feasibility of using the oncolytic properties of viruses to treat LM. The tested viruses included reovirus, poliovirus, and herpes simplex virus. In 2003, Wen Qing Yang and others experimented with the ability of human reovirus to kill medulloblastoma (MB) cell lines and surgical specimens in vitro and inhibit tumor growth/metastases in vivo [41]. Subsequently, it was found that intrathecal injection of reovirus significantly extended the survival of immunocompetent rats model with LM of breast cancer [42].

In 2004, Hidenobu Ochiai and others constructed a poliovirus (PV) recombinant oncolytic virus PVS-RIPO, achieved by genetic recombination with human rhinovirus type 2 (HRV2), and then explored its efficacy against breast cancer metastasis to the brain and LM in xenotransplantation rat models. The study found that intrathecal or intracerebral administration of PVS-RIPO was highly effective against human breast cancer xenografts growing in the subarachnoid space or brain parenchyma of athymic rats [43].

In 2023, Kuruppu and others published a paper on the use of conditionally replicating herpes simplex virus 1 (HSV1) to treat the growth of breast cancer LM at different stages (i.e., lag phase, log phase, and exponential phase). In the mouse model, these stages represent the early, middle, and late stages of LM in patients. Mice injected with oncolytic virus during the lag phase and log phase of LM had survival extended by 2 and 1 week, respectively, compared to the control group. This suggests that the therapeutic effect is related to the oncolytic rate of the virus and the growth rate of tumor cells. Although no complete remission was achieved in this experiment, tumor growth was inhibited as observed by Gd-MRI [44]. If the oncolytic rate is fixed, and the tumor growth rate is faster than the oncolysis rate, the disease manifests as continuous tumor growth. This may be the reason why there were no completely relieved mice in this study. There is still relatively little research on the mechanisms, and it is difficult to determine the real reason for this outcome.

Drug distribution and safety issues in early LM viral therapy

Drug distribution

Viruses are evenly distributed in the CSF through intraventricular or intrathecal administration. Driesse and colleagues injected recombinant adenoviruses containing the LacZ gene (encoding β-galactosidase enzyme, which can then hydrolyze X-gal into a blue product) into the CSF of non-human primates through intraventricular and suboccipital cisterna magna injections. After the application of X-Gal, widespread and intense blue staining of cells proved that the virus was evenly distributed along the meninges covering the brain and spinal cord, mainly in the arachnoid and pia mater [45]. Oshiro and colleagues assessed the distribution dynamics of retroviral vectors and their producer cells in rats and non-human primates, with uniform distribution of the virus after intrathecal and intraventricular injection [46]. In addition, Margarita and others evaluated the migration and distribution of intraventricularly injected neural stem cells (NSCs) in a mouse glioma model, with NSCs acting as viral transport vectors. The study proved that NSCs administered through the intraventricular route can effectively migrate to single or multiple tumor foci [47]. Therefore, intra-CSF administration is a very reliable administration strategy for OVT of LM.

Safety issues

In the aforementioned preclinical animal model studies applying viral gene therapy and viral oncolysis for LM, inflammatory reactions in the brain were observed, but no severe safety issues were reported.

In 1996, in a study using adenoviral gene therapy for LM in rats, the rats did not show any clinical or neurological signs related to the vector and/or GCV administration. Microscopically, no morphological or cytopathic effects were observed in the CNS [32]. This result was similar to an early study where a recombinant adenovirus was directly injected into brain tissue [48]. However, whether strong immune responses occur in the relatively "immune privileged" CNS after recombinant adenovirus injections is still a matter of debate [49,50,51].

In 2000, Driesse and colleagues compared the different responses of rats and primates to intra-CSF injection of recombinant adenoviruses. The results showed that rats exhibited dose-dependent clinical symptoms after administration, such as lethargy, conjunctival congestion, and weight loss. Neuropathological examination revealed multinucleated cell infiltration in the choroid plexus at the administration site; focal polymorphonuclear cell infiltration around the brain surface and spinal cord; and scattered red blood cell extravasation foci in the arachnoid. Macaques given the medication showed no clinical symptoms; however, an increase in IgG anti-adenovirus antibody titers in CSF and serum, and histopathological and immunohistochemical analysis of brain tissue revealed T-lymphocyte and plasma cell infiltration in the choroid plexus at the injection site, as well as mononuclear cell infiltration in the pia mater, indicating a combined cellular and humoral immune response to recombinant adenoviruses [52]. The CSF analysis and CNS histopathology of the macaques (3 weeks after viral injection) were consistent with viral meningitis [52]. Additionally, in 1999, Dewey and colleagues reported persistent brain inflammation 3 months after successful treatment of syngeneic gliomas in rats using the adeno-tk/GCV combination. The inflammatory infiltration was composed of activated macrophages/microglia and astrocytes, as well as T lymphocytes positive for leukocyte plasma protein, CD3, and CD8, and included secondary demyelination [53].

In 1996, in a study using HSV gene therapy for LM in rats, all rats treated with intrathecal injection of HSV vectors exhibited neurological symptoms 1–2 days after vector injection, characterized by increased lethargy and reduced intake of food and water. Control animals that received intrathecal injection of culture medium DMEM did not show any toxic symptoms at this early time point. The vector group rats with neurological onset recovered rapidly after the initiation of GCV treatment (day 2 after vector inoculation). However, animals in the vector group without GCV treatment showed prolonged morbidity of up to 7 days and some mouse deaths. This indicates that GCV can well control the replication of the virus carrying the HSV-tk enzyme. If the oncolytic virus is designed with certain enzymes that can convert certain drugs into harmful substances to itself, then this would be a good strategy to control viral shedding and rescue adverse reactions related to viral replication. Necropsy of the deceased animals in this experiment showed a small frontal dural focus and some small pial foci. LacZ staining indicated that the HSV vector was only present in tumor cells close to the CSF, and the tumor cell area showed necrosis and inflammatory infiltration. The researchers believe that death seems to be caused by oral and pial bleeding [34].

In 2004, in a study using reovirus OVT for LM in rats, local inflammation and mild communicating hydrocephalus were observed in animals treated with live-virus (LV), but not in dead-virus (DV) treated or control animals. In terms of local inflammation, calcification, local tissue reactions, and activated cellular reactions (microglia, macrophages, lymphocytes, etc.) were mostly observed around the injection site, with no significant "dose–response" when comparing 10^7 to 10^8 PFU. hematoxylin–eosin (H&E) staining and glial fibrillary acidic protein (GFAP) staining characterized that there was no diffuse inflammatory response found in the brains of live-virus and dead-virus treated groups. Morris water maze testing showed that all animals (i.e., normal group, DV group, and LV group) performed similarly, with no behavioral abnormalities found [41].

In 2011, in a clinical trial using retroviral vector-producing cells carrying HSV-tk to treat melanoma LM (GTI0108) [54], which is also the only phase I/II clinical study of viral therapy for LM, the first subject experienced severe toxicity, leading to the termination of the trial. The subject was infused with NIH3T3 fibroblasts (mouse-derived) capable of producing retroviral vectors expressing the thymidine kinase gene through intraventricular delivery, and 7 days later was treated with i.v. GCV for 14 days. The patient observed meningeal stimulation and high fever (40.6 °C) 5 min after the infusion of vector-producing cells, which was successfully managed with dexamethasone and tylenol. The presence of vector cells was detected by the expression of the NeoR marker gene in the CSF until the first dose of GCV on day 7, after which the copy number of the NeoR marker gene dropped to an undetectable level and remained so for 24 h. However, no significant changes in tumors were found in MRI scans or CSF cytological examinations before and after treatment. In this clinical study of intrathecal gene therapy for leptomeningeal cancer, GCV had a significant cytotoxic effect on vector-producing cells, but it was not enough to significantly reduce the extensive leptomeningeal tumor burden or the number of tumor cells in the CSF. The patient survived for 9 months after treatment, without autopsy. It is worth noting that the meningeal stimulation observed in the patient after the infusion of vector-producing cells was apparently caused by an acute inflammatory reaction of the meninges to the vector-producing cells, as NIH3T3 is mouse fibroblasts [17].

In a clinical trail published in 2021 on the use of neural stem cells to deliver recombinant oncolytic adenoviruses for intratumoral (i.t.) injection to treat gliomas (NCT03072134), one subject developed grade 3 viral meningitis due to an error that caused the vector to be injected into the CSF, which returned to normal after one week of antiviral treatment. Other patients in this study who received i.t. injection of the vector did not develop viral meningitis [55]. This seems to remind us the high possibility of viral meningitis caused by direct intrathecal injection of OVs. OVs delivered by neural stem cells do not necessarily avoid the occurrence of viral meningitis.

To sum up, the aforementioned exploratory studies using different viruses to treat different LM animal models have shown promising therapeutic effects. However, there are differences in inflammatory responses in the CNS and CSF during the treatment process. The CNS responds differently to different viruses, and even different recombinants of the same virus. When the same virus treats different LM animal models, the systemic symptoms and intracranial inflammatory responses presented by the interaction between animals and viruses are also different. Overall, OVT is generally safe, with a low incidence of adverse events, minor damage, and controllable or self-resolving issues [56]. However, unpredictable issues such as chronic meningitis, demyelination, and other long-term adverse events still require further experimental research. As clinical trials expand and more patients participate, more long-term or short-term adverse events may be reported and analyzed [17]. Therefore, when using viruses as vectors for LM therapy, the immune reactions mediated by the capsid and other possible adverse reactions must be considered to ensure safety by fully restoring potential adverse reactions in preclinical experiments. The higher fatal risk in the real world limits many developed OVs from being tested in clinical trials for patients with LM, necessitating the development of OVs that can be injected intrathecally or intraventricularly.

Intracranial immune status

The special immune environment in the CSF

Since the early nineteenth century, the brain has been considered an isolated, barrier-protected organ, thus possessing "immune privilege." This hypothesis was supported by experiments showing that dyes injected into the circulatory system did not stain the brain, and there was evidence that tissue grafts in the brain survived longer compared to those transplanted into peripheral tissues. However, over the past few decades, our understanding of the relationship between the brain and the immune system has completely changed. It is now established that immune cells act as guardians of the CNS, supporting brain function and repair, and reside in specialized immune niches at its borders, including the meninges, choroid plexus, and perivascular spaces [57]. The meningeal tissue is home to a large number of innate and adaptive immune cells [58, 59]. The lymphatic system in the meninges, CSF, and resident lymphocytes in the meninges form a relatively mature network, providing the physical conditions for antigens in the CSF to enter the lymphatic system through the cervical lymph nodes, thereby initiating the development and activation of T cells [60]. Recent results have found that microscopic channels in the skull allow bone marrow (BM) to be exposed to CSF, and CSF directly enters the regulatory skull marrow niche through dural channels, thereby affecting cell maturation and migration from BM to the brain [61,62,63]. A recent study by Huixin Xu and colleagues used real-time two-photon imaging technology to discover that during inflammation, the epithelial cells of the choroid plexus promote the destruction of the tight matrix that holds them together, creating an opening for immune cells to pass through, and it can be seen that neutrophils, monocytes, and macrophages enter the CSF from the blood, but another group of monocytes and macrophages can also enter the choroid plexus from the CSF in the opposite direction [64]. The OVT for LM relies on the special immune environment of the CNS to produce antiviral immune responses and subsequent anti-tumor immune responses. During pathogen invasion, antigen-presenting cells (APCs) located in the choroid plexus and meninges recognize and process viral antigens [65]. Viral antigens are transmitted to the cervical lymph nodes through lymphatic vessels, triggering specific T cell immunity [45]. A recent study by Jonathan Kipnis and others found that a large number of T cells are enriched in the anatomical structure of the dural sinus, which can also recognize antigens from the brain and CSF, thereby initiating an immune response [60].

Antiviral immunity and anti-tumor immunity in the CSF

After the injection of OVs into the CSF, the cranial immune system generates an immune response to the OVs and tumor cells, and the complex interactions ultimately achieve anti-tumor effects. There are mainly the following three mechanisms of action:

• I. Tumor-selective infection, replication, and oncolysis. The virus enters the cell through virus-specific receptors, taking advantage of the fact that tumor cells divide faster and have higher metabolism than quiescent cells, rapidly replicating progeny viruses, and after the replication is completed, the progeny viruses are released, and the tumor cells lyse and die. Modifying or mutating the surface proteins of the virus to recognize tumor-specific cell receptors can enhance the ability of OV to enter tumor cells. At the same time, the immune response function of tumor cells is dysfunctional, and the defective signaling of antiviral type I interferon (IFN) can hinder virus clearance [12, 66].

• II. Induction of systemic anti-tumor response. After viral replication, tumor cell death simultaneously changes from non-immunogenic to immunogenic (characterized by the release of DAMPs and tumor-associated antigens (TAAs)) mediating the body's anti-tumor immune response [67]. This process is called oncolytic virus-induced immunogenic cell death (ICD) of tumor cells, which can directly eliminate live tumors and lay the foundation for innate immune responses and adaptive anti-tumor immune responses. On the one hand, APCs located in the meninges and choroid plexus recognize OV infection, induce innate immune responses of NK cells and macrophages by secreting cytokines and chemokines [68, 69]. On the other hand, APCs initiate antigen-specific T cells by presenting viral antigens and tumor-associated antigens with costimulatory molecules to major histocompatibility complex (MHC) molecules, inducing anti-tumor specific CD8 + T cell immune responses. Once activated, naive CD8 + T cells become cytotoxic effector cells and are transported to the tumor site, where they mediate anti-tumor immunity, killing tumor cells that have not been infected by the virus and producing a lasting anti-tumor effect [67, 70, 71].

• III. Remodeling the immunologically "cold" tumor microenvironment (TME) into a "hot" one. The TME is a complex niche for the development of cancerous and noncancerous cell components, and it can be characterized as immunologically 'cold' or 'hot' depending on proinflammatory cytokine production and immune cell infiltration levels [72]. Immunotherapy based on OVs reshapes the "cold" characteristics of the tumor microenvironment, which are scarce in inflammatory immune cells, locally immunosuppressive, have poor prognosis, and are unresponsive to immunotherapy, into "hot" ones. A "hot" TME is associated with a higher response rate of activated immune cells, and tumors infected with OV are more easily recognized and attacked by the immune system [70, 72,73,74] (Fig. 1).

Fig. 1

Oncolytic viruses promote antitumor immunity. Oncolytic viruses reverse the immunosuppressive status of tumor cells and reduce the population of immunosuppressive cells, including T regulatory cells (Tregs) and M2 macrophages. The antiviral immune response within tumor cells recruits immune cells [e.g., CD4 + T cells, CD8 + T cells, innate lymphoid cells (ILCs), dendritic cells (DCs), M1 macrophages and natural killer (NK) cells], and activates them, converting "cold" tumors (tumors are tumors in a state of immune suppression) into "hot" tumors (tumors in a state of immune activation)

The antiviral immune response that coexists with the anti-tumor immune response should not be ignored, as it is one of the main causes of adverse reactions in the actual application of OVT. The antiviral immune response is mainly directed against viral proteins and is related to the immunogenicity of the virus [75]. It is independent of the virus infecting cells through chimeric antigen receptor (CAR) or integrins [76]. Viral replication or de novo protein synthesis does not seem to be a prerequisite for the immune response [77]. When OVs are administered intracavitary to treat other tumors (such as malignant pleural effusion, malignant ascites, ovarian cancer, etc.), it can lead to the reduction of cancerous effusions or ascites in patients [78]. However, unlike other body cavities, the cranial cavity is a fixed-volume bony cavity, and once an immune adverse reaction occurs during OVT for LM, it may lead to increased CSF, increased intracranial pressure, and even brain herniation and other serious adverse events [75] (Fig. 2). The occurrence of viral meningitis in early exploratory studies using viruses to treat LM has also been confirmed. As mentioned above, clinical symptoms such as somnolence, conjunctival congestion, and weight loss occurred, and in some studies, some experimental mice died. Neuropathological examination revealed T-lymphocyte and plasma cell infiltration in the choroid plexus at the administration site, focal polymorphonuclear cell infiltration around the brain surface and spinal cord; scattered red blood cell extravasation foci were observed in the arachnoid, and an increase in IgG anti-adenovirus antibody titers in CSF and serum indicated a combined cellular and humoral immune response to the virus in the CSF [32, 33, 41, 42, 52]. In addition, high concentrations of pro-inflammatory cytokines in the CSF of patients with infectious meningitis are associated with cognitive impairment and adverse outcomes [79, 80]. At the same time, the inflammatory process of the meninges cannot be separated from the inflammatory process of the brain. Animal model studies have shown that IFN-γ activation of the JAK-STAT pathway promotes the activation of microglia [81], changing the activity of neuronal networks by producing cytokines, reactive oxygen, and nitrogen. However, the co-activation of Toll-like receptor TLR4 and IFN-γ receptor leads to neuronal dysfunction and death, mainly caused by the enhanced expression of inducible nitric oxide synthase (iNOS) in microglia and the release of nitric oxide(NO) [82]. Therefore, due to the extensive distribution and mobility of CSF, viral meningitis caused by OVT for LM involves the entire brain and affects the entire cranial cavity, which may lead to serious consequences. However, when other body cavities such as the pleural cavity, abdominal cavity, bladder, etc., produce effusions due to OVT, patients do not face a fatal threat.

Fig. 2

The cranial cavity has a fixed volume. The presence of the skull means that changes in intracranial volume can lead to severe complications

As mentioned above, in the OVT for LM, the antiviral immune response is the main cause of viral meningitis. Reducing the antiviral immune response can reduce the incidence of related adverse reactions. However, whether reducing the antiviral immune response is beneficial to OVT is still controversial. In the OVT of solid tumors, on the one hand, the antiviral immune response can limit viral replication and spread, thereby reducing the direct oncolytic effect on cancer cells (and subsequently reducing the efficacy of OVT), so the innate and adaptive antiviral immune responses are considered to be detrimental to the efficacy of OV-based therapy. This seems reasonable [83, 84]. Therefore, many therapeutic strategies aimed at reducing the immune response against OV are currently being developed. For example, one method under study uses viral innate MHC inhibitors [85]. On the other hand, the antiviral immune response induced by OVT has significant anti-cancer benefits: the adjuvant-like characteristics of the antiviral innate immune response are absolutely crucial for the initial initiation of the anti-tumor immune response, and the antiviral immune events within the tumor transform the "cold" tumor into a "hot" tumor by promoting the recruitment of immune cells, overturning tumor-associated immunosuppression, and establishing a niche suitable for the development of anti-tumor immunity. Antiviral CD4+ helper T cells have the ability to shape the quality of anti-tumor CD4+ and CD8+ T cell immune responses, and the antiviral response targeting the site of viral replication also directly targets cancer cells because OV preferentially infects cancer cells [86]. Therefore, the antiviral immune response reduces part of the oncolytic effect in OVT, but at the same time, the antiviral immune response is beneficial for the body to produce adaptive anti-tumor immune responses and enhance anti-tumor effects. It is currently believed that appropriately weakening the host's antiviral immune response against OV is considered a strategy to promote the oncolytic activity of OV [84]. A correct understanding and utilization of the interaction between the antiviral immune response and the anti-tumor immune response may achieve a balance between the pursuit of higher efficacy and avoiding adverse reactions in OV-based therapy.

Potential biosafety issues of OVT for LM

Firstly, OVs may shed to other target organs. OVs are designed to specifically replicate in cancer cells, but this is only the most ideal state. I.t. injection can precisely control the concentration of OVs at the target site while preventing side effects caused by the virus mistakenly targeting other organs [87]. However, even so, viral shedding to other organs and causing adverse reactions is often observed in clinical practice. T-VEC(The first oncolytic virus approved by the U.S. Food and Drug Administration),which kills cancer cells through i.t. delivery, is mainly used for the treatment of melanoma. However, during T-VEC treatment, the virus may transfer to other parts of the patient's body or to people in close contact with the patient [87,88,89]. In the clinical trial of Ad5-124-RGD, Kimball and others found that patients receiving high doses of OVs frequently shed the virus, which could be detected in serum, urine, and saliva, most commonly in saliva, and the shedding rate seems to be related to the dose of oncolytic virus treatment [90]. Of course, the adverse reactions caused by viral shedding to other organs are related to the type of virus, the dose of the virus, and the virus's tropism for different organs. Viral shedding has been detected in treatments with HSV, adenovirus, vaccinia virus, and reovirus, while it is rarely observed in i.v. injections of vesicular stomatitis virus (VSV) or poliovirus [91]. Also, due to the hepatic and splenic tropism of adenovirus, some patients experience liver dysfunction, while other viruses rarely do [92,93,94,95]. Although viral shedding during the treatment process seems inevitable, a recent study review compared the incidence of major adverse events (WHO grades 3–4) in studies with proven therapeutic effects in cytotoxic drugs, targeted therapies, and immunotherapies, and found that the side effects of cancer vaccines and OVs on cancer patients are much lower than other systemic therapies [56]. Therefore, shedding to other target organs seems inevitable, but the adverse reactions caused by viral shedding are considered controllable.

Secondly, as previously studied using viral vectors for gene therapy of LM [45], viruses administered intrathecally circulate with the CSF throughout the entire cranial cavity, expanding the contact area between the virus and tumor cells. However, it also increases the contact area between the oncolytic virus and normal brain tissue, increasing the possibility of viral off-target effects. The leptomeninges, as the first line of defense of the brain-CSF barrier, are the first to be attacked. A recent preclinical study by Kang and colleagues inoculated γ134.5-deficient oncolytic HSV (G207) into the lateral ventricle of mice to determine the damage mechanism of intraventricular therapy and potential steps to reduce toxicity. They found that toxicity was caused by damage to the ependymal lining, partly due to viral replication and induction of CD8+ T cells [96]. In previous studies, phosphorylation of the protein kinase R that induces eukaryotic initiation factor 2α (eIF2α) by interferon (IFN) was not effectively initiated in ependymal cells, leading to toxicity [97]. They demonstrated through experiments using mouse weight loss as a toxicity indicator that pretreatment with a low dose of G207 or Poly I:C intraventricularly induced interferon phosphorylation of eIF2α can reduce ependymal toxicity. This pretreatment method can safely provide a variety of clinically relevant therapeutic doses of G207 and extend the survival of a disseminated brain tumor model [96]. This work has significant translational significance. G207 is the most widely studied oncolytic HSV in preclinical and clinical brain tumor research and has been proven to be safe in the CNS of adults and children [98]. Crucially, this preclinical study proves the safety/efficacy of intraventricular injection of G207, making people aware of the potential effective prospects of oncolytic virus therapy for LM. This makes G207 a potential candidate to be the first oncolytic virus to enter clinical trials for OVT of LM. However, more detailed safety and efficacy data of G207 intra-CSF injection in non-human primates are still needed.

In addition to infection, viral shedding may also lead to homologous recombination between OVs and residual wild-type viruses. When two similar viruses infect the same cell, the risk of homologous recombination is high, which may produce pathogenic transgenic viruses. This mechanism has not been observed in OVT, but it has been found in the production and use of vaccines [87, 99]. For example, T-VEC is an oncolytic virus drug developed from HSV-1, which can latently infect neurons, thereby inducing latent infections. Corrigan and others pointed out that the DNA of oncolytic HSV may persist in the neuronal bodies around the injection site and may induce severe HSV infections of the nervous system from a long-term perspective [100, 101]. In children with severe combined immunodeficiency (SCID) who received oncolytic retroviral therapy, viral gene integration into the LMO2 proto-oncogene region was found, leading to the development of leukemia and reducing the survival rate of these patients with impaired diseases [102]. However, different OVs treating different leptomeningeal metastatic cancers may have different toxicological reactions and biodistributions, and more basic research and preclinical animal studies are needed for a detailed understanding.

Administration strategies

Potential administration routes for OVT of LM include intra-CSF administration and systemic administration. Intra-CSF administration directly injects the drug into the CSF, bypassing the blood–brain barrier to enter the CNS. The advantage is that the drug can quickly reach effective concentrations, reducing systemic adverse reactions and enhancing drug action. The disadvantage is that it is more likely to cause adverse reactions such as viral meningitis. Meningitis patients usually exhibit fever, chills, abdominal pain, nausea, and headache [28, 103]. Other manifestations include shortness of breath, loss of appetite, neck stiffness and pain, and sensitivity to bright light [104]. In addition, inattention or double vision may also be related to meningitis and may persist after recovery [105]. Their occurrence mainly depends on the pathogenic virus [106].

Systemic administration of OVT for LM can deliver OVs throughout the body. Not only the primary lesion, but also some minor tumor lesions may be treated, including LM. The disadvantage is also obvious, systemic administration of OVs requires high selectivity for target tissues [107], otherwise, it may lead to systemic adverse reactions. Secondly, immune cells in the blood mediate more rapid immune clearance [108], although i.v. injection of oncolytic HSV is safe, there is some evidence that the virus replicates in extracranial solid tumors [109], but neutralizing antibodies, the blood–brain barrier, and first-pass metabolism, and other barriers hinder the virus from reaching the intracranial tumor in sufficient amounts to produce an adequate response. Preclinical trials have tried to overcome these challenges with various technologies, but they have not yet been tested in patients [110, 111].

In contrast, intraventricular or intrathecal vaccination bypasses many of these barriers, allowing more targeted administration. In addition to choosing which administration route to use, the dose of oncolytic virus particles is also a key factor in inducing the antiviral immune response. OVs need to reach a certain number to enter tumor cells, replicate, and then lyse tumor cells, releasing progeny viruses and repeating the infection of other tumor cells [112]. If there are too many virus particles, it will cause a strong antiviral immune response, leading to severe immune adverse consequences; if there are too few virus particles, it will lead to premature clearance by the immune system. In summary, choosing which delivery method is the key to the safety and efficacy of OVT for LM.

Methods to improve the biosafety of OVT

The ideal OV should meet several requirements: (a) high targeting or affinity for tumor cells, (b) rapid replication and amplification, (c) reduced antiviral response, and (d) activated immune response [84]. From the early viral gene therapy for LM, the distribution dynamics of the virus after intra-CSF administration were very good. However, due to the antiviral immune response and potential biosafety issues, the development of OVs for the treatment of LM remains a challenge [45]. We will discuss how to improve the safety and efficacy of OVs applied intrathecally from five aspects (Table 1): using safer OVs, reducing the immunogenicity of OVs, genetically engineering attenuated viruses, improving tumor cell targeting, and controlling viral replication.

Table 1 Strategies to enhance the biosafety of oncolytic viruses for the treatment of LM

Using safer OVs

To ensure biosafety, appropriately weakening the host's antiviral immune response against OV has been considered a strategy to promote OV oncolytic activity [87]. One method involves selecting viruses that are not infectious to normal tissues. Parvoviruses naturally infect rats and are non-pathogenic to humans, with low selectivity for non-malignant human cells. However, the overexpression of cytokines and transcription factors in tumors can activate metabolic pathways, regulating the function of non-structural protein 1 (NS1, an essential protein for viral DNA replication, gene expression, and virus-induced cytotoxicity), thereby increasing the tumor selectivity of parvoviruses [113, 114]. From cell culture, animal models to clinical application, PV has shown anticancer activity against various tumor types, such as melanoma, sarcoma, glioblastoma, neuroblastoma, Burkitt's lymphoma, liver cancer, breast cancer, lung cancer, gastric cancer, and pancreatic ductal adenocarcinoma. Compared to other OVs, PV is currently the smallest OV in clinical development, and its lack of an envelope makes it one of the few OVs that can cross the blood–brain barrier. Therefore, PV is a major candidate for treating brain tumors [115]. In 2015, a clinical phase I/II trial of H-1PV intervention for glioblastoma was completed. In this study of 18 patients, H-1PV was proven to be safe and well-tolerated. Compared with historical controls, this therapy improved progression-free survival and median overall survival, although the used protocol did not eradicate tumors [116]. The results are consistent with animal model studies, which predict that when administered systemically, the virus can cross the blood–brain barrier and induce immune transformation of the TME, including specific T cell responses [117]. Therefore, due to the selectivity of parvoviruses for multiple tumors and their clinically proven safety, choosing parvoviruses as OVs for the treatment of LM may treat LM while minimizing the potential biosafety issues of OVs. Because parvoviruses can cross the blood–brain barrier, even i.v. administration for the treatment of LM can be explored, which significantly avoids the strong antiviral immune response that may be caused by intra-CSF administration, thereby increasing safety. Although there are no related exploration studies at present, this is an interesting direction for research.

Reovirus is a naturally occurring oncolytic virus that is almost ubiquitous, and almost all human sera are positive for antibodies against reovirus [118, 119]. After reovirus invades normal cells, it can activate the double-stranded RNA-activated protein kinase R (PKR), which is a serine/threonine protein kinase that requires double-stranded RNA binding and phosphorylation to be activated. The main function of PKR is to protect normal cells from viral infection, promoting the anti-proliferative response of interferon after viral infection, so reovirus only exists in small amounts or cannot replicate in normal cells [120]. In tumor cells, reovirus uses the abnormal cell signal transduction /malignant phenotype produced by tumor cell mutations to preferentially replicate in tumor cells [121]. Therefore, it has the characteristics of preferentially replicating and lysing in tumor cells, is non-pathogenic or has low pathogenicity to the human body, and its pathogenicity to normal cells can be weakened by repeated passage [99, 122].

Virus packaging to reduce viral immunogenicity

In addition to using safer OVs, another method involves using special virus packaging to reduce the immunogenicity of the virus. Viruses with special packaging are manufactured using special techniques to create OVs with an outer surface coated with special materials, such as protein nanolayers, artificial liposomes, or red blood cell membrane-coated OVs. These specially packaged OVs avoid rapid clearance by the body after entering the human body, and the slower in vivo clearance rate allows the oncolytic virus to accumulate more in tumor tissue.

In 2019, Ran and colleagues published a paper stating that through the polyprotein surface precipitation (PSP) technique, polyproteins with opposite charges were sequentially adsorbed onto charged substrates to assemble nanospheres, and the remaining nanosphere structure could easily bind various materials [123]. Subsequently, Ran and others prepared adenovirus nanospheres with good biosecurity and reproducibility through a self-assembly method. The nanosphere technology improves the transfection efficiency of viral treatment, and the additional antigenicity of the entire nanosphere during in vivo transportation is minimized, which may hinder the immune response of T cells to the PSP technique. And the biological function of oncolytic adenovirus after release from the nanosphere in vivo is maintained [84, 124, 125].

In 2022, Huang and others published a paper for the first time exploring the ability of hybrid membrane vesicles of artificial lipid membranes and red blood cell membranes (red blood cell lipidosomes, RM-PL) to shield the antigens of oncolytic adenoviruses. They found that adding artificial lipid membranes (liposomes) to red blood cell membranes inherits the characteristics of both artificial and natural cell membranes, shielding the antigens of adenovirus 11 (ad11) better than liposomes or red blood cell membranes alone, resulting in reduced exposure of OV antigens, protecting ad11 from neutralizing antibody binding and macrophage uptake [126, 127]. At the same time, it extended the circulation of ad11 and avoided severe cytokine release syndrome (CRS) caused by i.v. injection of the virus, and the longer circulation further increased the accumulation of ad11 in tumors, enhancing the anti-tumor effect on primary and metastatic tumors. This hybrid membrane composed of artificial and natural cell membranes can overall coat antigen-rich and spike-like living biological particles to achieve complete shielding of OV antigens, not only avoiding the severe adverse reactions that the body's strong antiviral immune response may cause but also providing possibilities for various delivery methods of OVs [128].

In 2023, Huang and others published another paper, discovering that the serum clearance rate of AD11 viruses coated with polyethylene glycol liposomes (PEG-Lipo) nanoparticles was similar to that of bare AD11 viruses. Subsequently, it was found that the viral protein corona on the surfaces of bare ad11 and Lip-ad11 led to a rapid clearance of ad11 in the blood after i.v. injection. Therefore, they explored for the first time the plasma clearance rate of AD11 coated with artificial viral protein corona followed by PEG-Lipo. The results showed that this strategy extended the circulation time of OVs by more than 30 times and increased the distribution of OVs in tumors by more than 10 times [129]. The oncolytic virus treated with the artificial viral protein corona and liposome further concealed the viral antigens, and the in vivo retention time and tumor enrichment rate of the virus were proven to be increased in i.v. injection studies. This virus packaging strategy may reduce the viral meningitis caused by intrathecal injection of OVs for the treatment of LM.

Genetic engineering to construct attenuated OVs

Genetic modification has accelerated the development of OVT [130]. Researchers can construct desired viruses by deleting or inserting foreign genes according to needs to improve the safety, anti-tumor efficacy, or tumor targeting ability of OVT [91].

By genetically modifying the virus to delete some genes necessary for virus replication in normal cells or virulence genes, the safety of OVs can be improved [131]. The wild-type HSV-1 can cause latent infections in normal neurons, and its neurovirulence factor (ICP34.5) antagonizes the immune response mediated by protein kinase R (PKR) in normal cells, which is crucial for infecting neurons. Deleting its neurovirulence factor ICP34.5 can enhance its safety [132]. However, PKR is usually downregulated in tumor cells. Therefore, oncolytic HSV lacking ICP34.5 cannot replicate in neurons but can replicate in these tumor cells [133]. T-VEC and G207 are oncolytic HSVs that have deleted ICP34.5. As mentioned earlier, G207 has shown high safety in clinical trials for the treatment of gliomas in adults and children [134, 135]. The third-generation oncolytic HSV drug G47Δ further activates the body's immune system to clear tumors by deleting ICP-47 (which inhibits MHC-1 antigen presentation) based on G207 [136]. Oncolytic adenovirus H101 is the earliest approved oncolytic virus type for clinical application. By deleting the early replication-related genes of adenovirus (E1B, E3), it can prevent the replication of adenovirus in P53 function-normal cells but can replicate in commonly P53 function-abnormal tumor cells [137, 138]. Oncolytic adenovirus Onyx-015 is also designed with this strategy [139]. It is reported that the hepatotoxicity caused by adenovirus may be due to the combination of coagulation factor (F) X with the high-variable region (HVR) of the virus. Therefore, inserting mutations in the FX binding domain of HVR and replacing them with HVR from other serotypes of the original adenovirus can significantly reduce the liver tropism of oncolytic adenoviruses [140]. Other viruses, such as oncolytic vaccinia viruses, deleting genes such as TK, hemagglutinin, and B18R in their genomes can significantly reduce their virulence in normal cells [141, 142]. Therefore, more than 40 OVs are currently being tested, and most OVs have significantly improved their safety by genetically engineering to delete pathogenic genes, which is a very reliable strategy to enhance viral biosafety.

Another method is to recombine different types of viruses for treatment. The recombination of vesicular stomatitis virus (VSV) and Newcastle disease virus (NDV), named recombinant VSV-NDV (rVSV-NDV), greatly reduces the cytotoxicity to healthy hepatocytes and neurons and is non-pathogenic to embryonated eggs. In rVSV-NDV, the main chain of VSV is retained. However, its glycoprotein is replaced by hemagglutinin-neuraminidase (HN) and NDV envelope proteins. The off-target effects in the brain and liver observed in wild-type VSV experiments that cause adverse events are significantly reduced due to the replacement of glycoproteins [143]. Adenoviruses are widely used in recombinant OVT, such as adenovirus-coxsackievirus and adenovirus-parvovirus therapies [144]. Recombinant adenovirus-parvovirus retains the infectivity of adenoviruses and the harmlessness of parvoviruses to normal cells, thereby killing tumor cells and sparing normal cells [145]. Therefore, Recombinant viruses are also an excellent strategy to avoid off-target effects of viruses that may lead to brain tissue damage.

Enhancing the tumor cell targeting of OVs

Recently, Karan and his colleagues used a mathematical model of viral infection to determine the characteristics required for OVs to eradicate tumors without affecting non-cancer cells. The conclusion is that the virus must have a difference in its ability to infect these two different cell types, and the infection rate of non-cancer cells must be less than one percent of cancer cells. The difference in virus production rate or the death rate of infected cells alone is not enough to protect non-cancer cells [146]. Therefore, enhancing the targeting of OVs to tumor cells is also one of the measures to improve the safety of OVs.

By inserting targeting peptides, the ability of OVs to bind to viral receptors on the surface of tumor cells can be enhanced. The sequence peptide composed of arginine-glycine-aspartic acid (RGD) is widely found in proteins of the extracellular matrix (ECM) and can specifically bind to various integrins. Integrin αvβ3, which is highly expressed on the surface of various solid cancer tumor cells, can recognize the RGD peptide sequence [147]. By genetic engineering, OVs modified with targeting peptides can rely on highly expressed integrins to enter tumor cells, reducing the opportunity for the virus to enter normal cells [147]. DNX-2401 (Delta-24-RGD) is an oncolytic adenovirus modified by RGD, and DNX-2401 treatment has led to safe responses and significant responses in recurrent high-grade gliomas, as well as long-term survival [148]. RGD-modified Oncolytic Mammalian reovirus (MRV) also infects cancer cell lines that express junctional adhesion molecule A (JAM-A) and causes tumor cell death [149]. Meanwhile, RGD peptide-based delivery vectors, including RGD-based cationic polymers, lipids, peptides, and hybrid systems, are designed for cancer therapy. The aim is to particularly highlight the enhanced therapeutic effects and specific targeting ability exhibited by these vectors for cancer gene therapy both in vitro and in vivo [150]. There are other modifications that can enhance the tumor tropism of OVs. It is reported that the gene encoding a single-chain antibody (scAb) targeting human epidermal growth factor receptor 2 (HER2) was integrated into HSV-1, making it fully target tumor cells expressing HER2, while HER2-negative cells are unharmed, and this modification also enhances safety [151,152,153]. In 2015, Betancourt and others reported that they replaced the G protein of vesicular stomatitis virus (VSV) with the gp160 of human immunodeficiency virus type 1 (HIV-1), creating a new oncolytic virus (VSV-gp160G). This method abandons the natural tissue tropism of VSV and specifically targets the new receptor CD4. VSV-gp160G does not harm normal CD4 T cells; however, it shows strong cytotoxic activity against CD4-expressing tumor cells (such as adult T-cell leukemia/lymphoma cell lines), which is related to the defective antiviral immune pathway in these tumor cells [154].

In addition, using an external magnetic field to induce the specific distribution of OVs can also be used to overcome the potential safety risks associated with infection of non-tumor cells [155]. In 2015, Choi and colleagues published a paper on polyethylene glycol-coated magnetic iron oxide nanoparticles (PCION) to encapsulate oncolytic adenoviruses. Subsequently, they used an external magnetic field (EMF) to guide their migration, significantly improving the infectivity and specificity of oncolytic adenovirus to tumor cells in tumor-bearing mouse models [156].

Another method is to deliver OVs through cellular delivery. Common carriers for delivering OVs mainly include transformed cells, immune cells, and progenitor cells [157]. Several types of immune cells, including T cells [158, 159], monocytes/macrophages [160], and myeloid-derived suppressor cells have been studied as carrier cells because they can circulate throughout the body, have the ability to specifically recognize tumors or tumor-related characteristics, and have their own anti-tumor activity, thus carrying out one or two strikes on tumors. Progenitor cells used to deliver OVs include mesenchymal stem cells [161], neural stem cells, and vascular progenitor cells. They all have the ability to migrate to tumors [157]. Among them, neural stem cells have been specifically studied for the intracranial delivery of OVs to brain tumors. These cells are characterized by their ability to differentiate into nervous system cells (neurons, astrocytes, or oligodendrocytes) and their ability to self-renew [162]. A variety of clinical studies have used neural stem cells (NSCs) to deliver various OVs to treat brain tumors, including oncolytic adenoviruses and oncolytic herpes simplex viruses, and have shown potential therapeutic effects and biosafety [55, 163, 164]. It is worth noting that recently Margarita and others evaluated the migration and distribution of intraventricularly injected NSCs in a mouse glioma model. The study proved that NSCs administered through the intraventricular route can effectively migrate to single or multiple tumor foci. It reveals the potential of NSC-mediated oncolytic virus therapy for primary and metastatic brain tumors and LM [47].

Drug control of viral replication

Finally, in early gene therapy experiments for LM, the strategy was gene-directed enzyme prodrug therapy (GDEPT), which uses replication-deficient viruses to transduce foreign genes into cancer cells, increasing the selectivity of cancer cells to drugs to kill tumor cells [27]. This strategy can also be used when designing replicative OVs, thus the virus can be killed by giving the prodrug, controlling the death of the virus. This strategy combines the advantages of viral gene therapy and the oncolytic advantages of Ovs. the prodrug can kill the virus in advance when OVT causes a strong immune response, instead of waiting for the body clearing slowly. This strategy provides further protection for normal cells from viral damage.

In 1996, in a study using HSV gene therapy for LM in rats, rats treated with intrathecal injection of HSV viral gene vectors showed neurological symptoms 1–2 days after vector injection, but symptoms recovered rapidly after GCV treatment, while symptoms in mice not treated with GCV lasted for 7 days [34, 35]. GCV controlled viral replication well and alleviated clinical symptoms. In 2011, in a clinical trial using retroviral vector-producing cells carrying HSV-tk to treat melanoma LM (GTI0108) [54], the first patient was infused with NIH3T3 producer cells capable of producing retroviral vectors expressing the thymidine kinase gene through intraventricular delivery, and 7 days later was treated with i.v. GCV for 14 days. The presence of vector cells was detected by the expression of the NeoR marker gene in the CSF until the first dose of GCV on day 7, after which the copy number of the NeoR marker gene dropped to an undetectable level. The results showed that the trial was not sufficient to significantly reduce the extensive leptomeningeal tumor burden or the number of tumor cells in the CSF [17]. Although there is no experiment currently that clearly states whether this strategy will affect the anti-tumor effect of OVT, the rapid decline in the number of vector cells transporting the virus after GCV administration indicates that this strategy can be used to quickly kill the virus when the viral particles in the CSF are too high, to avoid subsequent potential damage to normal tissues.

Discussion

At present, clinical trials applying OVs in the cranial cavity often use stereotactic intra-tumoral injection to treat brain tumors, but there are no completed clinical studies using OVs to treat LM tumors. Research on OVT for LM is still in its infancy, with relatively few basic and preclinical studies.

The use of viruses to treat LM has gradually shifted from the initial viral gene therapy to OVT, with experimental viruses including retroviruses, adenoviruses, herpes simplex viruses, reoviruses, polioviruses, etc. Analysis of experiments with different types of viral treatments has found many similarities in the anti-tumor effects induced by gene therapy and OVT, and both therapies themselves have tumor-killing effects and can activate the body's anti-tumor immunity. Gene therapy kills tumor cells through prodrugs, while OVT kills tumor cells through the virus's replication in tumor cell and oncolysis.

Existing preclinical animal studies, whether viral gene therapy or OVT, have shown good therapeutic effects in treating LM. The efficacy is reflected in the extension of survival time after viral treatment, the extension of asymptomatic latency, the achievement of tumor-free long-term survival in animals, and the suppression of tumor growth monitored by technologies such as MRI. Overall, the efficacy of using viruses to treat LM in preclinical animal models is promising.

The early exploration of viral treatment for LM often used intraventricular or intrathecal administration, directly injecting into the CSF. It has been proven that this method of administration allows for uniform distribution of the virus, expanding the contact area between the virus and tumor cells, which is conducive to the virus playing a role. At the same time, it also expands the contact area between the virus and normal tissues, leading to adverse reactions. The most common adverse reaction is viral meningitis. When viral meningitis occurs, it often involves the entire cranial cavity, and widespread inflammatory factors can also cause brain damage. Once a severe immune adverse reaction occurs, it may lead to increased CSF, increased intracranial pressure, even brain herniation, death, and other serious adverse events. This is contrary to the original intention of treatment.

If systemic administration of OVs is chosen to treat LM, it can avoid the occurrence of severe viral meningitis and deliver OVs throughout the body including LM. The disadvantage is also obvious, i.v. administration of OVs requires high selectivity for target tissues, otherwise, immune cells in the blood will mediate more rapid immune clearance, and it may also lead to systemic adverse reactions. In contrast, intraventricular or intrathecal vaccination bypasses many of these obstacles, allowing more targeted administration. Therefore, intrathecal injection may still be the best route of administration for LM. However, although there are many OVs under development, the higher fatal risks in the real world limit many developed OVs from being tested in clinical trials for patients with LM. The immune environment in the CSF is special, and there is an urgent need to develop OVs that can be administered intrathecally.

To overcome the difficulty of viral meningitis after intra-CSF injection of OVs in LM therapy, we analyzed the feasibility of using viruses with low pathogenicity to humans, genetically modifying to reduce the immunogenicity of OVs, virus packaging to reduce the antiviral response of the immune system, enhancing the targeting of OVs, and drug control of viral replication, and other methods. The feasibility of these methods has been confirmed in other tumor studies and we speculate that it is also feasible in the OVT of LM. However, only the feasibility of drug control of viral replication has been verified in early gene therapy experiments for LM, which can quickly kill the virus to reduce the antiviral immune response in the CSF. Our argument is not unreasonable, but the specific performance of these strategies in LM therapy with OVT still requires a large number of preclinical experiments to verify from the use of viruses for gene therapy of LM to the current use of viruses for OVT, it shows people's continuous exploration of treatment strategies for LM. The exploration of new strategies of OVT for LM is still in its infancy. Early studies have shown the therapeutic value and also potential safety hazards. However, with the prosperity of OVT, people's in-depth understanding of viruses, scientific and technological progress, and the increasing incidence of LM, more and more researchers are focusing on the study of related mechanisms of action, and the development of OVs for the treatment of LM is worth looking forward to. In summary, OVT may be a very promising therapy in future treatment strategies for LM.