Introduction

Over 10% of cancer patients ultimately develop metastatic spinal cord compression. If left untreated, a tumor in the spine can quickly progress to paralyze the patient. Radiotherapy has long been used as an effective treatment for tumors involving both the spinal column and spinal cord. Yet this therapy carries a grave risk of causing myelopathy and irreversible neurological deficits. Because of this consideration, the conventional tolerance dose for spinal cord has been accepted as 45–50 Gy in 25 fractions. However, accumulated clinical experience indicates that the spinal cord may tolerate higher radiation doses. Recently the use of cutting edge technology has allowed the introduction of stereotactic radiosurgery to the treatment of spinal tumors, and this new approach has necessitated re-examination of the spinal cord tolerance in the context of radiosurgical and hypofractionated doses. With this platform of radiosurgery in humans, the partial volume tolerance dose has been established at 10 Gy to 10% of the ventral part of the spinal cord. This was found to be safe in clinical practice [1] and has been adopted in the RTOG 0631 clinical trial of spine radiosurgery [2].

The mechanism of radiation-induced myelopathy is not well understood. The classic radiobiological models have implicated parenchymal cell loss and vascular changes, yet a number of other mechanisms have been proposed as well. White matter necrosis may be one of the most common late histopathologic changes associated with irreversible neurological deficits [3]. More recent studies suggest a highly complicated pathogenic process involving vascular abnormalities, such as neovascularization, vessel wall thickening, vessel dilation, increased permeability and breakdown of the blood-spinal cord barrier [4]. Radiation-induced myelopathy is generally considered to be irreversible, and at present there are no established treatments available once the toxicity develops. In preclinical studies, one of the most promising approaches to attenuate late radiation injury involves inhibition of the metabolic pathway of the renin-angiotensin system (RAS). The precursor Angiotensin I is converted to Angiotensin II by angiotensin-converting enzyme (ACE). Angiotensin II signals through type 1 and type 2 receptors (AT1 and AT2). By using ACE inhibitors or blocking downstream angiotensin receptors, we and other investigators have shown mitigation of radiation-induced normal tissue toxicity in several organ systems including the optic nerve model, brain and kidney [5,6,7,8,9]. The mechanism of radiation mitigating effect remains to be fully understood, but it may also involve regulation of radiation-induced neuroinflammation [10]. In fact, recent evidence indicates that neuroinflammation by microglial activation may be an important contributing factor in the development of radiation-induced cognitive impairment [11]. In the present study, we hypothesized that treatment with the ACE inhibitor, Ramipril, would mitigate long-term radiation myelopathy, and that post-radiation neuroinflammation within the irradiated spinal cord may play a role. Using a rat model of radiation-induced spinal cord injury, we demonstrate that post-radiation Ramipril treatment results in functional motor recovery as well as microscopic structural restoration of the spinal cord after paralysis-inducing high doses of radiation. The Ramipril treatment also produces a significant reduction in microglial infiltration and vascular endothelial growth factor (VEGF) expression in the irradiated spinal cord tissue.

Methods

Animal subjects and treatment with Ramipril

Adult male Fisher 344 rats (Charles River Breeding Lab, Wilmington, MA) weighing between 200 and 240 g were used. All animal treatments were conducted in accordance with approved protocols reviewed by the institutional animal care committee of Henry Ford Hospital when an author (SR) was there. Rats were anesthetized with intraperitoneal injection (i.p.) of sodium pentobarbital (50 mg/kg) and immobilized using polystyrene foam. Single radiation doses of 23.6, 28.5, 30 or 33 Gy were delivered using a single collimated dorsoventral beam, centered to a 2-cm segment of the spinal cord between C4 and T2. The dose rate from a 6 MV linear accelerator was 2.5 Gy/min at 100 cm source to the surface distance. After irradiation with high doses (30 and 33 Gy), total of 22 rats were randomized to the Radiation only group (n = 12) and the Ramipril-treated (Radiation + Ramipril) group (n = 10). Sham control animals were also anesthetized but did not receive radiation nor Ramipril. Ramipril (Sigma; purity > 98%) was administered starting at 1 week after radiation to the spinal cord. Ramipril was given in the drinking water at a concentration of 1.5 mg/kg per day which was available at libitum. Ramipril treatment continued until the rats developed paralysis or until they were euthanized. Freshly prepared solution was given every day during the study period. The schema of the experimental design is shown in Fig. 1. Ramipril has highly predictable bioavailability, stability of the drug in water and its ability cross the blood brain barrier, and therefore blood-spinal cord barrier [12].

Fig. 1
figure 1

Schematic representation of the experimental protocol. A A single radiation dose of 23.6, 28.5, 30 or 33 Gy was delivered to a 2 cm segment of the spinal cord between C4 and T2. Ramipril was given in drinking water (1.5 mg/kg/day) starting 1 week after irradiation (RT + RAM). B Radiation field position. It was verified with a radiographic film

Full size image

Monitoring of neurological signs

The animals were monitored daily by the monitoring personnel of the animal facility who were blinded to the intervention for signs of neurological deficits and paralysis during the planned study duration of 6 months. For each rat, the study ended once the rats developed the first sign of paralysis involving one of both fore- and/or hind-limbs or were unable to support the full weight of the body. They were then euthanized according to the IACUC guidelines.

Histology and immunohistochemistry

After developing paralysis, the rats were anesthetized with an intraperitoneal injection of sodium pentobarbital (50 mg/kg) followed by intracardiac perfusion with PBS and 4% paraformaldehyde in PBS. The spinal cords were dissected and embedded in paraffin blocks. Each spinal cord was cross sectioned (6 μm thick) using a microtome and mounted on glass slides. Slides were deparaffinized in xylene and rehydrated through graded alcohol and distilled water. Adjacent serial paraffin sections were stained with hematoxylin and eosin (H&E) for histopathological assessment and Luxol Fast Blue/ Periodic Acid Schiff to evaluate myelination. To study the neuroinflammation with microglial infiltration and the endothelial and vascular effects, deparaffinized and rehydrated slides were heated in antigen unmasked solution (Vector, Burlingame, CA). Immunofluorescence staining was performed by overnight incubation at 4 °C with the following antibodies: rabbit polyclonal anti-Iba-1 (1:400; 019-19741, Wako Chemicals, Richmond, VA) and rabbit polyclonal anti-VEGFA (1:100; ab46154, Millipore, Temecula, CA). Secondary antibody against rabbit was incubated for 2 h at room temperature (Jackson, West Grove, PA). 4,6-Diamidino-2-phenylindole (Sigma, 1 µg/ml) was used for 15 min to counterstain nuclei.

Histological analysis and cell quantification

H&E sections spanning the region of spinal cord exposed to radiation were evaluated by a reader blinded to specimen identity. Grey and white matter were evaluated by light microscopy for the presence of vacuolation, hemorrhages and necrosis. Histological scores were assigned qualitatively, under low magnification, as mild (score 1) for histological changes observed in less than one third area of the spinal cord; moderate (scores 2–3) for histological changes between one third to one half of the spinal cord; severe (scores 4–5) for histological changes in more than one half of the spinal cord area.

For evaluation of demyelination, microscopic pictures of the spinal cord samples were taken after LFB stain using a Nikon Eclipse E600 Light Microscope equipped with a Nikon Digital Camera DSFi1. The total white matter area was calculated by subtracting the gray matter from the total spinal cord sectional areas using NIH Image J software. The percent demyelinated area was calculated as demyelinated area/total white matter area of spinal cord × 100.

For quantification of microglia stained with Iba-1 and cells expressing VEGFA, the positively stained cells were counted in three sections per animal for a minimum of six animals per group. For each section, representative confocal images of the dorsal funiculus were acquired at 40× magnification with a Leica TCS SP8X confocal microscope and cell counting was performed by randomly applying counting frames (200 × 200 µm). Cell counting results were converted into cells per 200 µm2 using Abercrombie’s correction formula [13].

Statistical analysis

Paralysis data were analyzed by Kaplan-Meier curves with significance determined by log-rank test. The number of demyelinated plaques per section and percentage of demyelinated area were analyzed by Student’s t test. VEGF and Iba1 positive cell counting were analyzed by one-way analysis of variance followed by Tukey’s post-hoc analysis. All statistical analyses were performed using GraphPad Prism 5 software. Differences were considered statistically significant at a p-value of <0.05.

We certify that all applicable institutional and governmental regulations concerning the ethical use of animals were followed during the course of this research.

Results

Ramipril treatment reduced the rate of paralysis and prolonged the time to develop paralysis after radiation to the spinal cord

To determine dose response and the ED100 for radiation-induced paralysis, single doses of 23.6, 28.5, 30 or 33 Gy were delivered to the rat spinal cord. There was a sharp dose response for developing paralysis between 23.6 Gy and 28.5 Gy during the 6-month duration of the study. None of the rats showed any signs of motor impairment after a single dose of 23.6 Gy irradiation. However, after the radiation doses of 28.5, 30 and 33 Gy, 100% of the rats treated with radiation (RT) developed limb paralysis at 135 days post-radiation (Fig. 2a). Since previous studies had shown that Ramipril treatment is an effective mitigator of radiation-induced optic neuropathy when administered early after radiation [9], we started the treatment with Ramipril 1 week after radiation to the spinal cord. Ramipril prolonged the latency to develop paralysis after a paralysis-inducing high radiation doses (30 and 33 Gy) to the spinal cord. The mean (±SD) time to develop paralysis after radiation was 125 ± 4 days in the RT group. In contrast, it was lengthened to 135 ± 4 days in RT + RAM group (P < 0.001) (Fig. 2b).

Fig. 2
figure 2

Radiation-induced paralysis attenuated by Ramipril treatment. A Rate of paralysis at 135 days after radiation to the spinal cord as a function of radiation dose. Each dot represents a single animal (n = 12). B Ambulatory probability curves show time to onset of paralysis after high radiation doses (30 and 33 Gy) in RT (red line, n = 12) and RT + RAM (blue line, n = 10). P < 0.001 by Log-rank test

Full size image

Ramipril treatment improved radiation-induced demyelination

Upon gross examination, the irradiated segment of the spinal cord of rats in the RT group showed a mild brown discoloration on the surface (arrows in Fig. 3a). The diameters of the spinal cords were not different among the experimental groups. Histology of H&E stained sections revealed a bubbly vacuolation in the grey and white matter of the spinal cord in the RT group (Fig. 3a). High magnification images showed hyalinization with areas of hemorrhage and extensive necrosis in the tissue of irradiated animals. These necrotic areas were associated with prominent vacuolization and axonal swelling. Thrombosed blood vessels were also identified (RT, Fig. 3b). Hemorrhage was present in the grey and white matter, but necrosis was observed only in the white matter. In the Ramipril-treated group, the tissue architecture was better preserved microscopically, and there was less vacuolization (RT + RAM, Figs. 3a, b). The hemorrhage and necrosis were less frequent in both the white and grey matter of Ramipril-treated spinal cords (RT + RAM, Fig. 3b).

Fig. 3
figure 3

Ramipril treatment improved radiation-induced demyelination. The animals received paralysis-inducing high radiation doses (30 and 33 Gy) and were sacrificed after paralysis. A Representative H&E stained sections from control (no radiation), RT and RT + RAM, 4× magnification. Insets show pictures of the gross spinal cords. Arrows indicate the region exposed to radiation in RT and RT + RAM animals. B H&E stains 40× magnification show large vacuoles and swollen axons in RT animals. Arrow point a degenerated and thrombosed blood vessel. Scale bar = 100 µm. C"?>c LFB stain. Intact myelin in control (no radiation). Arrows indicate areas of demyelination in sections from RT and RT + RAM animals. Scale bar = 200 µm. D Number of demyelinated plaques and the total area of demyelination per section in RT and RT + RAM animals. Data represent average ± SEM, n = 9 per group, **P < 0.01 by Student’s t test

Full size image

LFB stains showed restoration of white matter structural integrity in the Ramipril-treated spinal cord. In irradiated rats an average of two demyelinated plaques per section were identified in the lateral and dorsal column. Ramipril treatment showed a trend to decrease the mean (±SE) number of demyelinated plaques per section by 0.7 ± 0.6 (95% CI 0-1.4 plaques per section) (P > 0.05). On the other hand, the mean (±SE) area of demyelination was 6.7 ± 1.4% of the white matter per cross-section of the spinal cord in the RT group, whereas the demyelinated area was 0.2 ± 0.1% of the white matter in the RT + RAM group (P < 0.01) (Fig. 3c, d) with a mean between-group difference of 6.5% (95% CI 4.5%-8.5%).

Ramipril reduced microglial infiltration and VEGF expression after radiation to the spinal cord

We found that VEGFA expression was increased throughout the cross sections of the irradiated spinal cord, particularly in the dorsal column. There were also dilated capillaries with endothelium positively stained with anti-VEGF. The mean (±SE) number of VEGF positive cells was 4 ± 1 cells/200 µm2 in un-irradiated naïve controls. There was an 8-fold increase in the number of VEGF positive cells to 34 ± 2 cells/200 µm2 in the RT group (P < 0.001) with a mean between-group difference of 29 cells/200 µm2 (95% CI 24–35 cells/200 µm2). In contrast, Ramipril reduced the number of VEGF positive cells to 16 ± 1 cells/200 µm2 in the RT + RAM group (P < 0.001) (Fig. 4a, b). The mean difference between RT and RT + RAM was 17 cells/200 µm2 (95% CI 12-23 cells/200 µm2).

Fig. 4
figure 4

Ramipril treatment attenuated radiation-induced VEGF expression and Iba-1 expression. The animals received paralysis-inducing high radiation doses (30 and 33 Gy) and were sacrificed after paralysis. A and B Photomicrographs of VEGF stain and quantification of VEGF positive cells in the dorsal column. C and D Photomicrographs of Iba-1 immunostaining and quantification of Iba-1 positive cells in the dorsal column. Scale bar = 100 um. Insets in (c) show higher magnification image with microglia (63×). Data in (B) and (D) represent average ± SEM, n = 9 per group, *P < 0.05; ***P < 0.001 by ANOVA followed by Tukey’s post hoc test

Full size image

We evaluated infiltration of microglia/macrophages with Iba-1 staining in the dorsal column. The mean (±SE) number of Iba-1 positive cells was 2 ± 1 cells/200 µm2 in the un-irradiated naïve controls. There was a 3-fold increase in the number of Iba-1 positive cells to 7 ± 1 cells/200 µm2 in the RT group (P < 0.05) with a mean between-group difference of 5 cells/200 µm2 (95% CI 1–9 cells/200 µm2). Ramipril treatment significantly reduced the number of Iba-1 positive cells to 4 ± 1 cells/200 µm2 in the RT + RAM group (P < 0.05) (Fig. 4c, d). The mean difference between RT and RT + RAM was 3 cells/200 µm2 (95% CI 1–7 cells/200 µm2). Of note, the microscopic morphology of the microglia/macrophages from the RT group displayed an amoeboid shape suggesting an activated phenotype. In contrast, Iba-1 positive cells in the RT + RAM group had a more ramified shape suggesting the less activated phenotype similar to those of the unirradiated control group (Insets in Fig. 4c).

Discussion

To our knowledge this is the first study to demonstrate the mitigating effects of Ramipril in significantly reducing the paralysis rate and delaying the onset of paralysis after irradiation of the spinal cord with paralysis-inducing (ED100) dose. This mitigating effect was well correlated with inhibition of microglial infiltration and VEGF expression.

The dose response curve for radiation-induced paralysis has been well established in different animal models by either partial or full diameter irradiation of the spinal cord [14, 15]. Consistently with these results, our study revealed steep slope in the dose response curve for paralysis between 23.6 Gy and 28.5 Gy, 4 months after irradiation of 2 cm length of the full-diameter cervicothoracic spinal cord. Based on this finding, we chose the paralysis-inducing doses of 30 and 33 Gy to evaluate the effect of Ramipril in mitigating radiation myelopathy with paralysis as the clinical endpoint.

Multiple studies have demonstrated the mitigating effects of Ramipril for radiation injury to various organs including optic chiasm, hippocampus and whole brain, as well as kidney and skin. Of note, the mitigating effects were seen with different fractionation schedules mimicking a variety of clinical scenarios [7, 9, 16, 17]. Studies performed by Robbins et al. [18] and Moulder et al. [7] showed that hemodynamic changes in the kidneys after radiation could be normalized by ACE inhibitors, suggesting a role of Ang II in the pathogenesis of radiation damage to the kidney. ACE inhibitor treatment also protects the lungs from radiation-induced pneumonitis, fibrosis and pulmonary endothelial damage [16]. In the central nervous system, AT1 receptor antagonist L-158,809 and ACE inhibitor, Ramipril treatment ameliorated deficits in perirhinal cortex-dependent cognitive function after whole brain radiation to rats [5]. In a model of radiation-induced optic neuropathy, Ramipril treatment restored the gross and microscopic structural integrity of myelination of the optic nerves. It also resulted in functional improvement as assessed with improvement of the visual evoked potentials [8, 9]. Various mechanisms have been studied in the RAS metabolic pathway, but the exact mechanism of Ramipril mitigation is not clearly understood. Nevertheless, a recent study demonstrated cognitive improvement with complete elimination of microglia by blocking colony-stimulating factor-1 receptor [11]. Consistently with this finding, in our study the mitigating effect of Ramipril for radiation spinal cord damage was correlated with decreased microglia infiltration. Taken together these results strongly suggest a role for neuroinflammation in the pathogenesis of spinal cord radiation damage. Importantly, the results raise further questions about the temporal relationship of microglial activation during the dynamic process of radiation damage. Further studies are needed to answer the questions of when microglial infiltration/activation begins, where the cells come from (i.e., resident microglia vs. infiltrating macrophages), and the precise role that neuroinflammation plays in mediating radiation-induced paralysis.

Several lines of evidence suggest potential mechanisms for Ramipril’s ability to mitigate neuroinflammation. Direct infusion of Ang II to the brain increases adhesion of leukocytes to pial venules, expression of ICAM-1, VCAM and E- and P-selectin adhesion molecules and blood-brain-barrier disruption [19, 20]. In experimental models of atherosclerosis and hypertension, Ang II enhanced the expression of chemokines, MCP-1 and MIP-1α resulting in monocyte/macrophage infiltration, and it was ameliorated by ACE inhibitors and AT1 antagonists [21]. On the other hand, Ang II blockade has been shown to suppress microglia activation after LPS stimulation [22]. Ang II stimulates these cellular responses by inducing the expression and DNA-binding capacity of the transcription factor nuclear factor-κB (NF-κB) which controls many genes associated with inflammation [23].

It has been shown that radiation causes endothelial damage and decrease in vascular density with subsequent hypoxia, leading to VEGF upregulation [24]. VEGF is a potent endothelial mitogen and permeability factor that induces fenestrations in the endothelium of small venules and capillaries [25, 26] and the expression of adhesion molecules like ICAM-1 [27] further compromising the integrity of the blood-spinal cord barrier. Ang II induces VEGF mRNA production in vascular smooth muscle cells and heart endothelial cells that could be prevented by AT1 blockade [28]. In experimental models of diabetic retinopathy, ACE inhibitors and AT1 antagonists prevent the overexpression of VEGF and hyperpermeability as well as retinal neovascularization [29]. Upregulation of VEGF could be detected at 15 weeks after radiation to the rat spinal cord and precedes the appearance of demyelination and necrosis in the white matter at 18–20 weeks after radiation [24]. Our study demonstrated that Ramipril significantly reduced the VEGF expression, and this was also correlated with the improved outcome of paralysis. The results suggest that Ang II may mediate radiation-induced vascular damage and may also have contributed to the migration of the inflammatory cells to the injured area of the spinal cord through VEGF-induced changes in capillary permeability.

This study is limited by the use of the high paralysis-inducing radiation doses which are far above the radiation doses used for patients in the clinic. In extrapolating the experimental data for clinical application one must be further cautioned to take into account factors such as patient comorbidities and interactions with concomitant systemic therapies. Nonetheless, Ramipril is a well-tolerated, FDA approved drug that crosses the blood brain barrier [12] and therefore also crosses blood-spinal cord barrier. In conclusion, our study supports the translational potential for using Ramipril as a mitigator of radiation-induced normal tissue toxicity in patients.