Only very few studies on late effects in normal tissue are presently available [
11,
13,
23,
24]. Radiation-induced myelopathy is a feared late side effect in the CNS, characterized by a long symptom-free latency period followed by a sudden occurrence of neurological symptoms. To prevent the development of these severe complications, specific tolerance doses have to be respected and due to the uncertainty in the knowledge of the RBE, this is associated with significantly larger uncertainties for carbon ions than for photons.
To investigate the accuracy of RBE-predictions by the LEM, a large-scale dose–response study in the rat spinal cord has been performed. This animal model is well established for the investigation of late effects in the CNS and has been previously used to study the effectiveness of different beam modalities [
25‐
30]. Especially, it has been shown that the response of the spinal cord is independent of the irradiated volume for field lengths above 8 mm [
31,
32]. The model is also well-suited to study the temporal development of radiation-induced myelopathy in MRI as well as on the histological level. This study currently presents the largest and most systematic data base.
Dose–response studies
The rat spinal cord was used to characterize the RBE-variation along the central axis of a 6 cm SOBP for different fractionation schedules. The details of these studies have been published previously [
14‐
16]. Detailed in vivo testing of the RBE-predictions of LEM I and IV as a function of LET and fractional dose revealed that the RBE in the high-LET region is better described by LEM IV while the predictions of LEM I are more accurate in the low-LET region. It has to be noted, however, that this result refers to relatively high fractional doses. An additional dose–response study with 6 Fx is currently under evaluation and will allow to extend the benchmarking of the LEM also towards lower doses per fraction. Together with the presented results, this study will allow to estimate the α/β-value, which represents the extent of tissue regeneration in fractionated treatments. Preliminary results based on the single and split dose studies suggest an increase of α/β with increasing LET, indicating a decreasing impact of fractionation for increasing LET. For a more reliable estimation, however, the 6 Fx study has to be included. It has to be emphasized that the benchmarking of RBE-models is not restricted to the LEM. Currently, tests are extended to the Microdosimetric Kinetic Model (MKM) which is used for carbon ion therapy at National Institute of Radiological Science (NIRS, [
33,
34]).
MRI-based longitudinal study
The MRI-based longitudinal study enables a non-invasive investigation of occurring radiation-induced effects during the symptom free latency time. We found a fixed sequence of alterations in the images. Comparing the carbon ion and photon irradiations at isoeffective doses with respect to the endpoint paresis grade II, the same morphological changes were found and the only difference was a shorter latency time after carbon ion irradiation. Main findings in MRI were presence of edema, syrinx, uptake of contrast agent due to the break-down of the BSCB and finally followed by paresis grade I and II. Once the edema occurred in an animal, it developed the deterministic sequence. These findings were also confirmed quantitatively by evaluation of the DCE-measurements, which showed that the increase of the extracellular volume, ve, and the contrast agent exchange rate, Ktrans, increased similarly for carbon ions and photons.
It appears likely that the shorter latency time after carbon ion irradiations originates from differential actions on the histological or molecular level and apparently, MRI at 1.5 T is not sensitive enough for the detection of such alterations. With respect to sensitivity, the small diameter of the rat spinal cord and the consequently occurring partial volume effects may also play a role. Using an MRI with higher field strength would in principle be an option to increase the sensitivity, yet, in the present study, this was logistically not possible due to the excessive number of measurements, which had to be performed on a short term notice during the period where neurological symptoms appear within a rapid time sequence.
Despite of these limitations, this study provides the first extensive temporal characterization of the development of radiation-induced myelopathy after irradiation with carbon ions and photons in MRI and in an ongoing MRI-based histological study, tissue samples at different time points after irradiation as well as at the occurrence of the different endpoints in MRI are acquired. By investigating these samples on the histological and molecular level, more detailed information on the underlying mechanistic processes is expected.
Molecular mechanisms and inhibition
Currently, it is not clear in detail whether the target structures of irradiation in the spinal cord are the neurons or the blood vessels. Therefore, many attempts have been made to evaluate the effects of ionizing radiation to the neuronal [
22,
35‐
37] and the vascular proportion [
11,
24,
38‐
41] supporting nowadays the view that endothelial cells are the main target structure [
42‐
44].
At the endpoint paresis grade II, histological examinations revealed a comparable breakdown of the tissue structure for both radiation modalities; however, the increase of blood vessel permeability was much higher after carbon ion irradiation. This finding is in contrast to results of the DCE evaluation, where no difference was seen at the same endpoint.
It has to be noted, however, that increased permeability of the BSCB was detected with albumin, which presents a much larger molecule than MRI contrast agent Gd-DTPA (66 vs. 0.5 kDa). The discrepancy between the results of MRI and histological analysis could therefore be explained by a different extent of perforation for the two irradiation modalities. While the higher ionization density of carbon ions introduces more complex, non-reparable DNA-damage, which leads to intense blood vessel perforation and thus to an increased permeability for Gd-DTPA as well as for albumin, photons exhibit a low ionization density which induces better reparable DNA-damage and leads only to small vessel perforations and thus to an increased permeability for Gd-DTPA but much less for albumin. To clarify this, additional histological investigations with smaller molecular markers are required.
Besides vascular changes, also a profound damage of the neuronal structures was observed. Luxol fast blue staining shows a clear decrease of the myelin basic protein at the biological endpoint paresis grade II. To assess the relative importance of vascular and neuronal damage, a detailed investigation of the temporal development of both structures on the histological and molecular level will be performed within the ongoing MRI-based histological study.
Detailed knowledge of the mechanistic processes may enable targeted pharmacological interventions with the aim of protecting the normal central nervous system tissue after irradiation. First attempts along this direction have already been described in the literature [
45‐
48] using ACE-inhibitors. Within a pilot trial, we used the ACE-inhibitor ramipril™ to test the impact on radiation-induced myelopathy after carbon ion and photon irradiation. The rationale for using this drug are manifold: ramipril™ has been shown to exhibit mitigative properties on optic neuropathy [
47,
49]. In addition, with regard to the central nervous system, the drug is able to cross the blood-spinal cord barrier [
50], does not reveal protective effects on tumors [
51] and is already used to treat hypertension in patients. Our results showed that myelopathy could not be prevented, however a prolongation of latency time was achieved, which indicates that ramipril™ has a mitigative effect in the rat spinal cord. Identification of the underlying pathological pathways leading to radiation-induced side effects would facilitate the application of appropriate protective drugs and, if successfully realized, could allow elevating the tumor dose without harming the surrounding normal tissue.