Targeted therapy of spinal cord glioma with a genetically modified Salmonella typhimurium

Abstract

Objective:  Spinal cord tumours are highly malignant and often lead to paralysis and death due to their infiltrative nature, high recurrence rate and limited treatment options. In this study, we measured antitumour efficacy of the Salmonella typhimurium A1‐R tumour‐targeting bacterium strain, administered systemically or intrathecally, to spinal cord cancer in orthotopic mouse models.

Materials and methods:  Tumour fragments of U87‐RFP were implanted by surgical orthotopic implantation into the dorsal site of the spinal cord. Five and 10 days after transplantation, eight mice in each group were treated with A1‐R (2 × 10⁷ CFU/200 μL i.v. injection or 2 × 10⁶ CFU/10 μL intrathecal injection).

Results:  Untreated mice showed progressive paralysis beginning at day 6 after tumour transplantation and developed complete paralysis between 18 and 25 days. Mice treated i.v. with A1‐R had onset of paralysis at approximately 11 days and at 30 days; five mice developed complete paralysis, while the other three mice had partial paralysis. Mice treated by intrathecal injection of A1‐R had onset of paralysis at approximately 18 days and one mouse was still not paralysed at day 30. Only one mouse developed complete paralysis at day 30 in this group. Intrathecally treated animals had a significantly better survival than the i.v. treated group as well as over the control group.

Conclusions:  These results suggest that S. typhimurium A1‐R monotherapy can effectively treat spinal cord glioma.

Introduction

Spinal cord tumours are highly malignant and often lead to paralysis and death (1). In humans, intramedullary spinal cord tumours (IMSCTs), important among spinal cord tumours, are treated by surgical resection, radiation and chemotherapy. However, prognosis of IMSCT, especially for high‐grade glioma, remains poor (2) mainly due to their infiltrative nature, high recurrence rate and limited treatment options (3, 4, 5). IMSCTs usually remain asymptomatic when they are small and may increase to a considerable size before they are detected. Despite their gross total resection, residual microscopic disease can remain in the resection bed because of the progressive infiltrative behaviour of this cancer. With radical surgery and adjuvant therapy, 2‐year survival rates for patients with high‐grade glioma range only from 0% to 40% as reported in some studies (6, 7). Therefore, novel approaches to treatment of IMSCT are needed.

Coley observed more than a century ago that some cancer patients became cured of their tumours following post‐operative bacterial infection (8). In the middle part of the last century, Malmgren et al. showed that anaerobic bacteria had the ability to survive and replicate in necrotic tumour tissue with low oxygen content (9). Several approaches aimed at utilizing bacteria for cancer therapy have subsequently been described (10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22). Bifidobacterium longum has been shown to grow selectively in hypoxic regions of tumours after intravenous administration. This effect was demonstrated by Yazawa et al. in 7,12‐dimethylbenzanthracene‐induced rat mammary tumours (21, 22). Vogelstein et al. created a strain of Clostridium novyi, an obligate anaerobe, which was depleted of its lethal toxin (23); this strain of C. novyi was termed C. novyi NT. After intravenous administration, C. novyi NT spores germinated in avascular regions of tumours in mice, causing damage to surrounding viable tumour zones, but did not eradicate the tumours (23). Combined with conventional chemotherapy or radiotherapy, intravenous C. novyi NT spores caused extensive tumour death within 24 h (23).

After attenuation by purine and other auxotrophic mutations, the facultative anaerobe Salmonella typhimurium has been used for cancer therapy (18, 24, 25). These genetically modified bacteria replicated in tumours to levels more than 1000‐fold higher than in normal tissue (18). The msbB mutant of S. typhimurium was found to cause less septic shock (26) than others. To raise the therapeutic index, S. typhimurium was further attenuated by deletion of the purI and msbB genes (26). The new strain of S. typhimurium, termed VNP20009, could then be safely administered to patients in a Phase I trial involving metastatic melanoma and metastatic renal cell carcinoma (26). More studies are needed to completely characterize the safety and efficacy of these tumour‐targeting bacteria.

Mengesha et al. utilized S. typhimurium as a vector for gene delivery by developing a hypoxia‐inducible promoter (HIP‐1) to limit gene expression to hypoxic tumours. HIP‐1 was able to drive gene expression in bacteria residing in human tumour xenografts implanted in nude mice (27). Genes linked to the HIP‐1 promoter showed selective expression in tumours (27). Yu et al. used green fluorescent protein (GFP)‐labelled bacteria to visualize tumour‐targeting abilities of three pathogens: Vibrio cholerae, S. typhimurium and Listeria monocytogenes (28, 29).

Our group developed a strain of S. typhimurium, termed A1, which selectively grew in human tumours in nude mice. In contrast, normal tissue rapidly cleared infecting bacteria, even in immunodeficient athymic nude mice. S. typhimurium A1 is auxotrophic (leu/arg‐dependent), but receives sufficient nutritional support from tumour tissue. Previously A1 has been shown to cause PC‐3 tumour growth inhibition and regression as subcutaneous xenografts in nude mice (30).

To increase tumour‐targeting capability of S. typhimurium A1, the strain was re‐isolated after infection into a human colon tumour, growing in nude mice. The tumour‐isolated strain, termed A1‐R, had higher tumour‐targeting ability both in vitro and in vivo. S. typhimurium A1‐R demonstrated efficacy in treatment of orthotopic mouse models of human breast cancer (31) and orthotopic mouse models of human prostate cancer (32).

In the present study, we demonstrate that S. typhimurium A1‐R, administered systemically or intrathecally, can effectively treat human spinal cord cancer in a subcutaneous as well as in an orthotopic nude mouse model, suggesting the clinical potential of this approach.

Materials and methods

RFP vector production

The (DsRed‐2) gene (BD Biosciences Clontech, Palo Alto, CA, USA), was inserted in the retroviral‐based mammalian expression vector pLNCX (BD Biosciences Clontech) to form the pLNCX DsRed‐2 vector. Production of retrovirus resulted from transfection of pLNCX DsRed‐2 into PT67 packaging cells, which produced retroviral supernatants containing DsRed‐2. Briefly, PT67 cells were grown as monolayers in DMEM supplemented with 10% FCS (Gemini Biological Products, Calabasas, CA, USA). Exponentially proliferating cells (in 10‐cm dishes) were transfected with 10 μg expression vector using LipofectAMINE Plus (Life Technologies, Grand Island, NY, USA). Transfected cells were re‐plated 48 h after transfection and 100 μg/mL G418 was added 7 h later. Two days after, medium was changed and the level of the G418 selective agent was increased to 200 μg/mL. After 25 days of drug selection, surviving colonies were visualized by fluorescence microscopy and RFP‐positive colonies were isolated. Several PT67‐RFP clones were selected and expanded into cell lines after virus titering on the NIH 3T3 cells (33, 34, 35).

RFP gene transduction of tumour cells

For RFP gene transduction, 70% confluent U87 human glioma cells were incubated with a 1:1 precipitated mixture of retroviral supernatants of PT67‐RFP cells, in RPMI 1640 or other culture media (Life Technologies) containing 10% foetal bovine serum (Gemini Biological Products) for 72 h; fresh medium was replenished at this time. Tumour cells were harvested using trypsin/EDTA and subcultured at a ratio of 1:15 into selective medium, which contained 50 μg/mL G418. To select brightly fluorescent cells, the level of G418 was increased to 800 μg/mL in a stepwise manner. Clones of U87 expressing RFP were isolated with the use of cloning cylinders (Bel‐Art Products, Pequannock, NJ, USA) using trypsin/EDTA, U87 clones were amplified and transferred by conventional culture methods in the absence of selective agent (33, 34, 35).

GFP gene transfection of S. typhimurium

Salmonella typhimurium (ATCC 14028) was grown at 37 °C to midlogarithmic phase in liquid Luria Bertani (LB) and harvested at 4 °C. Bacteria (2.0 × 10⁸), in 40 μL of 10% glycerol, were mixed with 2 μL of pEGFP vector (Clontech, Mountain View, CA, USA). The preparation was placed on ice for 5 min before electroporation, using the Gene Pulser apparatus (Bio‐Rad, Hercules, CA, USA) according to the manufacturer’s instructions. Electroporation was carried out at 1.8 kV with the pulse controller at 1000‐Ω parallel resistance (36).

Induction of bacterial mutations with nitrosoguanidine and selection for auxotrophs of  S. typhimurium‐GFP

Freshly prepared nitrosoguanidine (NTG; 1 mg/mL in sterile water) was added to the washed culture to a final concentration of 100 μg/mL in Tris–maleic acid buffer (pH 6.0). S. typhimurium‐GFP were incubated with NTG for 30 min. NTG‐treated cells were grown in nutrient broth to express any mutations that had been induced. Bacterial colonies were replica‐plated in supplemented minimal agar plates containing specific amino acids, to identify auxotrophic requirements. Auxotroph A1, which required leu and arg, was initially isolated (30).

Selection of a high‐tumour‐targeting variant S. typhimurium A1

Selection of a high‐tumour‐targeting variant of S. typhimurium A1‐GFP was carried out as follows: A1 bacteria were injected into the tail vein of an HT‐29 human colon tumour‐bearing nude mouse. Three days after infection, tumour tissue was removed from the infected animal. Tumour tissue was then homogenized and diluted with phosphate‐buffered saline (PBS). The resulting supernatant of the tumour tissue was cultured in LB agar plates at 37 °C, overnight. The bacterial colony with the brightest green fluorescence was selected and cultured in 5 mL of LB medium. This strain was termed A1‐R (31).

Growth of bacteria

A1‐R bacteria were grown overnight in LB medium and then diluted 1:10 in LB medium. Bacteria were harvested at late‐log phase, washed with PBS and then diluted in PBS. Bacteria were then ready for injection into mice (31).

Subcutaneous and orthotopic transplantation of red fluorescent protein‐expressing U87 human glioma cells

Female nu/nu nude mice (4–6 weeks old) were used as host for the U87‐RFP human glioma cell line. The mice were anaesthetized with a ketamine mixture (10 μL ketamine HCl, 7.6 μL xylazine, 2.4 μL acepromazine maleate and 10 μL H₂O) before surgery, via s.c. injection. RFP‐expressing U87 cells (1 × 10⁶) were injected into subcutaneous tissue using a 0.5 mL 28 gauge latex‐free syringe (TYCO Health Group LP, Mansfield, MA, USA). For subcutaneous tumour models, tumour fragments resulting from cell injection (5 mm × 5 mm × 5 mm) stably expressing RFP, were implanted subcutaneously through a 5 mm incision on the lower back of the mice (37).

For the orthotopic IMSCT model, tumour fragments (0.5 mm), harvested from subcutaneously growing tumours, were implanted by surgical orthotopic implantation into the spinal cord. A midline incision approximately 2 cm long was made over the midthoracic spine. Subperiosteal dissection of the paravertebral muscles was then performed. Spinous processes and bilateral lamina at the midthoracic level (T‐7) were removed using a blade to expose the dura mater. A 28 gauge needle was inserted into the dorsal centre of the spinal cord, avoiding blood vessel injury, to create a 1 mm longitudinal incision. The U87‐RFP tumour fragment was implanted into the incision. Tumour pieces were determined to be stably expressing RFP. The muscles, fascia and skin were closed with a 6‐0 surgical suture. After recovery, the animals were returned to their cages (37).

S. typhimurium A1‐R killing of glioma cells in vitro

U87 human glioma cells, labelled with RFP, were grown in 24‐well tissue culture plates to a density of 1 × 10⁴ cells/well. A1‐R bacteria expressing GFP were grown in LB and added to the tumour cells (1 × 10⁵ CFU/well). After 15 min incubation at 37 °C, observation was initiated for 3 weeks.

Bacterial therapy in the subcutaneous tumour model Determination of subcutaneous tumour size

Two weeks after subcutaneous transplantation of tumour fragments into 30 mice, 14 mice in which the long diameter of the tumour was from 7 to 12 mm were selected for this experiment. Seven mice (treatment group) were administered S. typhimurium A1‐R (2 × 10⁷ CFU/200 μL PBS) intravenously once a week for 3 weeks. The remaining mice (control group) were administered the same volume of PBS. After administration of S. typhimurium A1‐R, fluorescence imaging was performed and changes in the diameters of the RFP‐expressing tumors were recorded each week for 3 weeks using the iBox Imaging System (UVP LLC, Upland, CA, USA).

Tumour diameters were measured each week after S. typhimurium A1‐R administration. Tumour volume was calculated by the formula:

Tumour volume in each group was expressed as mean ± SE. Statistical analysis was performed using the two‐tailed Student’s t‐test.

Bacterial therapy in the orthotopic IMSCT model

Five and 10 days after transplantation of tumours, mice were treated with S. typhimurium A1‐R (2 × 10⁷ CFU/200 μL i.v. injection or 2 × 10⁶ CFU/10 μL intrathecal injection). For intrathecal injection, mice were anaesthetized and placed on a sterile area. The prominent L7 spinous process was identified through palpation of iliac crests and a 0.5 cm longitudinal incision was made over the dorsal lower‐lumber region. Underlying fascia was swept laterally and the spinous process at L7 and ligamentum flavum were removed, exposing the intervertebral space. A 33‐gauge 1/2‐inch removable needle connected to a 10 μL syringe (Hamilton, Reno, NV, USA) was inserted through the dorsal L6–L7 intervertebral space and then S. typhimurium A1‐R was injected. Eight mice were treated with S. typhimurium A1‐R by i.v. injection; eight mice via intrathecal injection; and eight mice were used as the untreated control group.

Functional evaluation of hind limbs to determine degree of paralysis

Functional evaluation of hind limb strength was assessed using the Basso, Bresnahan and Beattie (BBB) scale (38). Mice were placed in an open field testing area and were observed for 5 min. Locomotion was rated using the BBB locomotor scale which ranges from 21 to 0 (21 means consistent plantar stepping and coordinated gait, consistent toe clearance, predominant paw position is parallel throughout stance, consistent trunk stability and tail consistently up. Zero means no observable hind limb movement) (Table 1). All animals were tested pre‐operatively to ensure a baseline locomotor rating of 21. After tumour transplantation, the animals were tested three times a week. Two different observers were randomly assigned to score motor function of the animals. Results of BBB score are expressed as means. The experiment was concluded by day 30. Dead animals in each group were recorded with a zero (0) functional score.

Table 1.

 BBB Functional score of hind limb function

Scale range	Range of movement
0	No observable hind limb (HL) movement
1	Slight movement of one or two joints, usually the hip and/or knee
2	Extensive movement of one joint or extensive movement of one joint and slight movement of one other joint
3	Extensive movement of two joints
4	Slight movement of all three joints of the HL
5	Slight movement of two joints and extensive movement of the third
6	Extensive movement of two joints and slight movement of the third
7	Extensive movement of all three joints of the HL
8	Sweeping with no weight support or plantar placement of the paw with no weight support
9	Plantar placement of the paw with weight support in stance only or occasional, frequent, or consistent weight‐supported dorsal stepping and no plantar stepping
10	Occasional weight‐supported plantar steps; no FL–HL coordination
11	Frequent‐to‐consistent weight‐supported plantar steps and no FL–HL coordination
12	Frequent‐to‐consistent weight‐supported plantar steps and occasional FL–HL coordination
13	Frequent to consistent weight‐supported plantar steps and frequent FL–HL coordination
14	Consistent weight‐supported plantar steps, consistent FL–HL coordination, and predominant paw position during locomotion is rotated when it makes initial contact with the surface as well as just before it is lifted off at the end of stance; or frequent plantar stepping, consistent FL–HL coordination and occasional dorsal stepping
15	Consistent plantar stepping and consistent FL–HL coordination and no toe clearance or occasional toe clearance during forward limb advancement; predominant paw position is parallel to the body at initial contact
16	Consistent plantar stepping and consistent FL–HL coordination during gait and toe clearance occurs frequently during forward limb advancement; predominant paw position is parallel at initial contact and rotated at lift off
17	Consistent plantar stepping and consistent FL–HL coordination during gait and toe clearance occurs frequently during forward limb advancement; predominant paw position is parallel at initial contact and lift off
18	Consistent plantar stepping and consistent FL–HL coordination during gait and toe clearance occurs consistently during forward limb advancement; predominant paw position is parallel at initial contact and rotated at lift off
19	Consistent plantar stepping and consistent FL–HL coordination during gait, toe clearance occurs consistently during forward limb advancement, predominant paw position is parallel at initial contact and lift off, and tail is down part or all of the time
20	Consistent plantar stepping and consistent coordinated gait, consistent toe clearance, predominant paw position is parallel at initial contact and lift off, and trunk instability; tail consistently up
21	Consistent plantar stepping and coordinated gait, consistent toe clearance, predominant paw position is parallel throughout stance, and consistent trunk stability; tail consistently up

Functional evaluation of hind limb function was assessed using the Basso, Bresnahan and Beattie (BBB) scale (38). The BBB scale ranges from 21 to 0 (21 means consistent plantar stepping and coordinated gait, consistent toe clearance, predominant paw position is parallel throughout stance, consistent trunk stability and tail consistently up. Zero means no observable hind limb movement). After tumour transplantation, the mice were tested three times a week.

Survival analysis

Kaplan–Meier curves were used to demonstrate survival distribution. The log‐rank test was used to compare survival data between the two groups.

Results and discussion

Efficacy of S. typhimurium A1‐R on human glioma cells in vitro

The ability of S. typhimurium A1‐R to kill human glioma cells in vitro was tested first. A1‐R expressing GFP was observed to invade U87 human glioma cells expressing RFP, almost immediately. Bacterial infection led to rapid death of a few cancer cells within 30 min of infection, and to death of most cancer cells within 120 min (data not shown).

Wild‐type S. typhimurium also kills tumour cells in vitro. However, wild‐type Salmonella kills mice rapidly, therefore we did not test it in vitro since we could not use it in vivo (30). We have previously compared A1‐R with A1 (31) and showed that A1‐R is much more effective in killing cancer cells.

Efficacy of  S. typhimurium A1‐R therapy on human gliomas growing subcutaneously in nude mice

The efficacy of systemic administration of S. typhimurium A1‐R on development of U87 subcutaneous glioma tumours, in nude mice is shown in Fig. 1. Average tumour size on day 0 (the day S. typhimurium A1‐R was administered) was 467 mm³ in the treatment group and 464 mm³ in the control group. In the treatment group, on day 7, tumour size was 689 mm³; on day 14, 665 mm³ and on day 21, 815 mm³. In the control group, tumour size on day 7 was 1406 mm³; on day 14, 3629 mm³ and on day 21, 5410 mm³. At days 14 and 21 after S. typhimurium A1‐R administration, tumour size in the treated group was significantly smaller than those in the control group (Fig. 1).

Figure 1.

  Efficacy of Salmonella typhimurium A1‐R therapy on U87‐RFP glioma subcutaneous tumour growth. (a) Imaging of tumour growth in the control group. (b) Imaging of tumour growth in the treatment group. Mice in the treatment group were given S. typhimurium A1‐R weekly i.v. for 3 weeks, beginning 14 days after tumour transplantation (day 0). Tumour size was measured using RFP imaging at day 0, 7, 14, and 21. C: Tumour volume of A1‐R‐treated mice and untreated mice was compared. Tumour volume in the treated group was significantly smaller than in the untreated group (at day 7, P = 0.05; at day 14: P = 0.009; and at day 21: P = 0.006).

Efficacy of  S. typhimurium A1‐R therapy on hind‐limb paralysis in the orthotopic IMSCT model

The efficacy of S. typhimurium A1‐R on hind‐limb paralysis in the U87 glioma IMSCT orthotopic model is shown in Fig. 2. Untreated mice experienced onset of progressive paralysis beginning at 6 days after U87 glioma transplantation, and developed complete paralysis between 15 and 25 days after tumour transplantation. Mice treated with S. typhimurium A1‐R via intravenous injection had onset of paralysis at approximately 11 days after transplantation. At day 30, five mice developed complete paralysis, while the other three mice had partial paralysis. Most of the mice treated by intrathecal injection had onset of paralysis at approximately 15 days after transplantation and one mouse was still not paralysed at day 30. Only one mouse developed complete paralysis at day 30 in this group. There was a significant delay in complete paralysis because of bacterial treatment (log‐rank test; control group and intravenous group: P < 0.05, control and intrathecal: P < 0.001, intravenous and intrathecal: P < 0.05).

Figure 2.

  Efficacy of Salmonella typhimurium A1‐R therapy on hind‐limb paralysis in the IMSCT orthotopic model. (a) The spinal cord of an untreated control mouse that was completely paralysed was exposed by laminectomy. RFP‐expressing tumour growing in the spinal cord was observed under fluorescent light. The primary tumour and a skip metastasis (arrow) were observed. (b) Average mean BBB score in each group. Five and 10 days after tumour transplantation in the spinal cord, eight mice were treated i.v. with S. typhimurium A1‐R (2 × 10⁷ CFU/200 μL); and eight mice were treated via intrathecal injection (2 × 10⁶ CFU/10 μL). The experiment was concluded on day 30. Results of BBB score are expressed as a mean. Mice in the control group demonstrated progressive paralysis starting at day 6 and eventually developed complete paralysis within 25 days. In the i.v. treated group, most of the mice had onset of paralysis by day 13. Five mice developed complete paralysis and the other mice had partial paralysis at day 30. Seven of the mice treated intrathecally with S. typhimurium A1‐R had onset of partial paralysis at day 25 and one mouse was still not paralysed at day 30. Only one mouse developed complete paralysis at day 30 in this group.

The most important objective in the treatment of spinal cord cancer is to prevent or delay progression of paralysis and increase eventual survival. Therefore, paralysis and survival as end points were chosen. Suppression of tumour growth should strongly correlate with the delay or prevention of paralysis.

Survival efficacy of  S. typhimurium A1‐R therapy in the orthotopic IMSCT model

Survival efficacy of S. typhimurium A1‐R therapy in the IMSCT orthotopic model is shown in Fig. 3. The U87 IMSCT orthotopic model resulted in rapid death of untreated animals. Of eight untreated mice with orthotopic U87 IMSCT, seven died within 4 weeks of tumour implantation. However, weekly bacterial treatment prolonged survival of tumour‐bearing mice. Of the eight mice with orthotopic U87 IMSCT, treated i.v. with S. typhimurium A1‐R, three were still alive on day 30. Of the eight mice with orthotopic U87 IMSCT, treated intrathecally with S. typhimurium A1‐R, seven were alive on day 30. There were significant differences between the intrathecal injection group and the control group (P < 0.005) and between the intrathecal‐injection group and intravenously‐injected group (P < 0.05). Paralysis resulting from the spinal cord glioma was in the hind legs rather than in the breathing muscles. Survival was assumed to be related to additional variables such as distant metastasis.

Figure 3.

  Survival efficacy of Salmonella typhimurium A1‐R therapy in the orthotopic IMSCT model. Seven out of eight untreated mice with orthotopic U87 IMSCT died within 4 weeks of tumour implantation. Of the eight mice treated i.v. with A1‐R, three were alive at day 30 and seven were alive in the intrathecally treated group at day 30. Using Kaplan–Meier analysis, there were significant differences between the intrathecal injection group and the control group (P < 0.005) and between the intrathecal injection group and the i.v. group (P < 0.05).

Glioma has multiple genetic and chromosomal abnormalities, which cause these tumours to be highly aggressive. The tumours invade rapidly, infiltrate and destroy neighbouring tissue, especially in the spinal cord. Current treatment of IMSCT involves radical or subtotal tumour resection by adjuvant radiation and/or chemotherapy, depending on the tumour’s histological type and grade as well as the extent of resection (3, 39, 40, 41). Limitations of these treatments include inadequate surgical resection, possibility of radiation‐induced damage by radiotherapy and limited efficacy of various chemotherapy regimens. Because of its poor prognosis, development of a new treatment strategy for IMSCT is urgently needed.

It has been demonstrated in this report that bacterial monotherapy effectively treated high‐grade IMSCT. Treatment using S. typhimurium A1‐R, a facultative anaerobe, can eradicate high‐grade IMSCT without surgery and/or toxic chemotherapy. Our approach to bacterial therapy of cancer is a major advance over the use of obligate anaerobes, such as Clostridia, which require combination chemotherapy to effect tumour eradication (42, 43).

It was also reported in this study that intrathecal administration of S. typhimurium A1‐R was a more effective route than i.v. treatment. Intrathecal treatment can overcome the blood–brain barrier, which significantly limits penetration of systemic chemotherapeutic agents, and thereby delivers a lower, less toxic dose of drug. Intrathecal chemotherapy is usually administered through a ventricular reservoir or lumbar puncture and has been available in the clinic for the last 40 years. Indeed, it is both widely and successfully used to prevent or treat central nervous system tumour spread, in haematological malignancies. However, there remains concern that because of disturbed cerebral spinal fluid dynamics, intrathecal treatment might be ineffective or unsafe (44). However, recent experience in the use of intrathecal chemotherapy for medulloblastoma has been encouraging. Success of the German HIT‐SKK 92 trial, using intrathecal methotrexate instead of radiotherapy for medulloblastoma, has helped to raise awareness of the potential efficacy of intrathecal chemotherapy, while avoiding radiotherapy, in very young children (45).

In the present study, intrathecal administration of bacteria as a therapy was tested in mice. Weekly administration of bacteria by intrathecal injection not only had no apparent toxicity, (which would have resulted in conditions, such as meningitis) but also resulted in prolonged survival and late onset of hind‐limb paralysis in animals of the U87 orthotopic IMSCT model.

Both A1 and A1‐R are auxotrophic for both leucine and arginine, which appears to preclude these bacteria from mounting a sustained infection in normal tissue. Tumours, however, appear to be able to supply these amino acids sufficiently and enable A1 and A1‐R to grow‐in and kill tumours (30). A1‐R was selected for increased virulence by passage through a growing tumour in a mouse (31). Other aspects of tumour selectivity of A1 and increased tumour selectivity of A1‐R remain to be elucidated.

In conclusion, we have demonstrated that S. typhimurium A1‐R monotherapy rapidly kills human U87 glioma cells in vitro. In the mouse subcutaneous tumour model, U87 tumour cells were highly sensitive to S. typhimurium A1‐R therapy. In the orthotopic IMSCT mouse model, S. typhimurium A1‐R, administered systemically or intrathecally, prolonged survival of the mice and strongly inhibited the progression of hind‐limb paralysis. Intrathecal administration of S. typhimurium A1‐R was a more effective route than intravenous administration, in the spinal cord cancer orthotopic model. Bacterial therapy of IMSCT is a novel and effective treatment strategy for spinal cord tumours, which can avoid surgical resection and radiotherapy, that cause damage to the spinal cord.