- Email: [email protected]
Dynamics of MRI lesion development in an animal model of viral-induced acute progressive CNS demyelination
www.elsevier.com/locate/ynimg NeuroImage 21 (2004) 576 – 582
Dynamics of MRI lesion development in an animal model of viral-induced acute progressive CNS demyelination Istvan Pirko, a Jeff Gamez, b Aaron J. Johnson, b Slobodan I. Macura, c,d and Moses Rodriguez a,b,* a
Department of Neurology, Mayo Clinic, Rochester, MN, USA Department of Immunology, Mayo Clinic, Rochester, MN, USA c Department of Biochemistry, Mayo Clinic, Rochester, MN, USA d NMR Core Facility, Mayo Clinic, Rochester, MN, USA b
Received 21 May 2003; revised 4 August 2003; accepted 15 September 2003
Theiler’s murine encephalitis virus (TMEV) infection in mice is an established model of CNS demyelinating diseases. The aim of the study was to determine the chronological pattern of lesion development in this model of monophasic fulminant demyelinating disease. We followed six highly susceptible interferon-gamma receptor knockout mice with serial in vivo brain magnetic resonance imaging (MRI) studies to determine changes in overall T2 lesion load and gadolinium enhancement. Altogether, 163 individual lesions were followed over 52 days. The number of lesions increased linearly with time. Four chronological patterns of lesion development were seen: (a) expanding lesions (48.5% of all lesions, 54.05% volume contribution); (b) expanding – retracting lesions (20.85% of all lesions, 15.03% volume contribution); (c) fluctuating lesions (16.6% of all lesions, 28.8% volume contribution); (d) stable lesions (14.05% of all lesions, 2.12% volume contribution). Gadolinium enhancement was not seen in the evolution of every lesion. Enhancement was both time- and lesion type-dependent. Early in the disease course ( < 43 days after infection), enhancement was almost always seen, later on (>43 days after infection) it was only seen in 8% of new lesions. All of fluctuating, 85.3% of expanding, 83.5% of expanding – retracting, and 56.5% of stable lesions were associated with gadolinium enhancement. We conclude that the MRI features of TMEV-induced demyelination in this model showed four unique chronological patterns, and inconsistent gadolinium enhancement. These novel findings may provide new insights into the pathogenesis of acute fulminant multiple sclerosis (MS). D 2003 Elsevier Inc. All rights reserved. Keywords: Lesion development; CNS demyelination; MRI
Introduction Multiple sclerosis (MS) is the most common inflammatory demyelinating disease of the central nervous system (Noseworthy et al., 2000). The disease has important socioeconomic impact in that
* Corresponding author. Department of Neurology, Mayo Clinic, Rochester, MN 55905. Fax: +1-507-284-1086. E-mail address: [email protected] (M. Rodriguez). Available online on ScienceDirect (www.sciencedirect.com.) 1053-8119/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2003.09.037
it is the leading cause of disability in young adults in developed countries. Magnetic resonance imaging (MRI) is a well-established tool in diagnosing and monitoring MS patients (Thorpe et al., 1994; Young et al., 1981). In addition, it has also become a frequently used secondary outcome measure in many MS-related therapeutic clinical trials (Barkhof and van Walderveen, 1999). The role of MRI in MS has increased given the newly established McDonald’s criteria, which puts greater emphasis on MRI studies (McDonald et al., 2001). The dynamics of lesion formation in MS has been studied (van Waesberghe et al., 1998; van Walderveen et al., 1998). T2-weighted sequences are especially sensitive to structural alterations of CNS tissue. All stages of MS lesion formation, from early inflammatory to late chronic, are visualized as hyperintense regions on T2weighted scans. Active lesions can be distinguished from chronic lesions by their increased permeability of the blood – brain barrier, which is demonstrated as enhancement of the gadolinium contrast on T1-weighted scans. With the advent of in vivo microscopic resolution small animal MRI, similar monitoring has become available for frequently used animal models of MS, including Theiler’s murine Encephalitis Virus (TMEV) infection. TMEV (Dal Canto et al., 1995) is a picornavirus that normally causes gastrointestinal infection in mice, occasionally accompanied by encephalitis. However, if the virus is inoculated intracerebrally, then a biphasic disease develops in susceptible strains of mice. The first phase lasts approximately 2 weeks. This phase is characterized by encephalitis and meningitis. In the second phase (beginning within 3 weeks after infection), immune-mediated demyelination is observed, which is similar to human demyelinating disorders (Brahic, 2002; Brahic and Bureau, 1998). Even though this model has been extensively studied since 1934 (Brahic and Bureau, 1998), its MRI features have not been characterized. The goal of our study was to characterize MRI lesion development in the TMEV model of MS in interferon-gamma receptor knockout mice (IFN-gR / ). We chose to do our MRI investigations in these mice because they develop an acute progressive demyelinating illness that results in death within 6 – 8 weeks, due to extensive brain and brainstem lesions (Johnson et al., 2001; Murray et al., 2002; Rodriguez et al., 1995). This is in contrast to
I. Pirko et al. / NeuroImage 21 (2004) 576–582
the disease that develops in the most commonly studied SJL/J strain. SJL/J mice develop a chronic demyelinating illness that is almost exclusively limited to the spinal cord, with little or no brain pathology observed. We studied in IFN-gR / mice the significance of gadolinium enhancement and the dynamics of T2 lesion formation by three-dimensional microscopic resolution in vivo magnetic resonance imaging at eight time points during the disease course.
577
the right parietal cortex. Intracerebral injection results in greater than 98% incidence of demyelination in mice of susceptible haplotype with only a rare fatality following injection. All animals develop mild viral encephalitis, which resolves within 14 days after injection. The animals then develop a chronic demyelinating disease in the brain and spinal cord that progresses over months. Magnetic resonance imaging (MRI)
Materials and methods Mice 129-Ifngrtm1 (IFN-gR / ) mice were generously provided to us by Michel Aguet (Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland). Animals were housed and bred in Mayo’s animal care facility. All animal studies conformed to the guidelines for use of animals by the Mayo Clinic and the National Institutes of Health. Theiler’s virus infection TMEV-induced demyelination was produced by intracerebral virus injection of 4 – 8-week-old mice anesthetized with methohexital. With a 27-gauge needle attached to a Hamilton syringe, a 10-Al volume containing 200,000 plaque-forming units (PFU) purified Daniel’s strain of TMEV was injected intracerebrally in
Six IFN-gR / mice were followed with high resolution 3D MRI scans on days 4, 10, 20, 27, 35, 43, 47, and 52 after TMEV infection. The choice of time points was in part determined by scanner availability. One animal expired on day 35. All five remaining animals expired on or after day 52 but before the planned subsequent imaging time point. The total scan time for all experiments was 136 h. The MRI examinations were performed in a Bruker Avance 300 MHz (7 T) vertical bore NMR spectrometer equipped with ‘‘miniimaging’’ accessories. The 37jC core temperature of the animals was maintained by a thermocouple-based system. Inhalational isofluran anesthesia (1.5% in oxygen) was delivered via a nose cone during the imaging procedure. Following the i.v. injection of ‘‘triple-dose’’ Magnevist (0.2 ml/100 g body weight of 1:3 physiologic saline-diluted Magnevist solution) T1-weighted volume-acquisition spin-echo sequences were used to visualize gadolinium enhancing lesions (TR: 400 ms, TE: 6.2 ms, FOV: 4 2.5 2.5 cm, matrix: 256 160 160). T2-
Fig. 1. Example of lesion development in an IFN-gR / mouse infected with TMEV. (A) Day 4, no lesions. (B) Day 10, four lesions; (C) Day 20, eight lesions; (D) Day 27, 14 lesions; (E) Day 35, 21 lesions; (F) Day 43, 29 lesions; (G) Day 47, 34 lesions; (H) Day 52, 36 lesions. The images represent 3D volume rendering resampled from T2-weighted 3D datasets of the studied animals. Lesions were extracted as detailed in Materials and methods, and object maps were generated to describe the lesions. Every lesion was color coded individually to help identification.
578
I. Pirko et al. / NeuroImage 21 (2004) 576–582
weighted imaging was performed using a 3D RARE volume acquisition sequence (TR: 1500 ms, TE: 70 ms, Flipback: ON, RARE factor: 16, FOV: 3.5 3.5 3.5 cm, matrix: 160 160 160). Analysis of 3D image sets: brain extraction, coregistration, segmentation, and volumetry Analyze 4.0 and 5.0, developed by the Biomedical Imaging Resource at the Mayo Clinic, was used for processing and analysis of the obtained 3D image sets. The datasets were segmented, and the brains were extracted using the Object Extractor and Image Edit tools. To correct for potential misalignment of the extracted brains at different time points, a 3D surface matching algorithm was used. This algorithm generates a rotation matrix. The rotated and coregistered 3D brain images were resampled with windowed sync interpolation of the original images. The co-registered images were then segmented using the 3D ROI Analysis Tool. A semiautomated intensity-based seed-growing algorithm was used to generate object maps. The object maps defined sub-volumes of the brain. Each subvolume is a stand-alone, numbered, individually identified, threedimensional brain lesion. The generated object maps were then applied to the subsequent time point, and were corrected as appropriate with the same seed-growing algorithm so that changes
in lesion number, volume, and shape could be accounted for. Because we used semiautomated tools, there was a human factor involved in this analysis. To minimize this, the involved investigators were trained for several weeks on test datasets, and their intra- and inter-rater reliability was repeatedly tested and was found to be superior (>95%). After the generation of object maps describing each time point in every animal, the 3D ROI Scan Tool was used to calculate individual and total lesion volumes.
Results MRI lesion development as a function of time Altogether, the animals developed 163 lesions. The number of lesions increased linearly in every animal (Figs. 1 and 2). There was no significant difference between the numbers of lesions among the animals at the studied time points. The total lesion volume per animal increased in a biphasic fashion. There was an initial linear increase until day 43; afterwards, a more rapid expansion of total lesion load was seen (Fig. 3). No significant difference was observed among lesion loads of animals at the last time point before death.
Fig. 2. Number of lesions per animal. The number of lesions increased linearly in all studied animals. The figure shows data of all animals studied. The mean values are also presented.
I. Pirko et al. / NeuroImage 21 (2004) 576–582
579
Fig. 3. Changes in lesion load (total lesion volume per time point) for each animal. At every represented time point, the total volume of all observed lesions was determined for each animal. The curves showed a biphasic growth of total lesion load with an initial linear phase, which was followed by a phase of rapid increase after day 42. The mean lesion volumes are also presented.
All 163 lesions were followed individually. None of the lesions observed on T2-weighted scans or gadolinium-enhanced T1weighted scans resolved during the illness. MRI datasets visualized on 3D movies To visually demonstrate the development of lesions in this model, 3D movies were generated using MRI datasets obtained from one mouse. Movie 1 shows a 3D rotation of a mouse brain extracted from the in vivo dataset. Every full rotation represents a new time point. The brain is rendered as a semitransparent object to allow for visualization of the individually color-coded lesions. Movie 2 is a time-lapse movie: the same semitransparent brain is seen from above; the developing lesions are color coded individually with the same method.
and expanding – retracting lesions (21.76 F 15.27). The mean volume of stable lesions was the smallest (6.17 F 2.51). Expanding lesions represented 54.05% of the total lesion load at the last time point before morbidity, followed by fluctuating (28.8%) and expanding – retracting lesions (15.03%). The total volume of stable lesions contributed least to the total lesion load, and altogether represented only 2.12% of the combined total lesion load of all animals at the last time point (Fig. 5). Gadolinium enhancement was not an absolute prerequisite for lesion formation Gadolinium enhancement was not seen in a consistent fashion during the formation of new lesions. Fluctuating lesions (n = 27) always showed this phenomenon in the first 3 – 7 days of devel-
Four chronological patterns of lesion development were observed
Table 1 Main characteristics of lesion types
The lesion formation as observed on T2-weighted scans followed one of four patterns (Table 1). From most common to least common, these were (a) expanding (continuous growth, n = 79, 48.5% of all lesions), (b) expanding – retracting (growth followed by partial resolution, n = 34, 20.85%), (c) fluctuating (growth – retraction – growth, n = 27, 16.6%), or (d) stable lesions (n = 23, 14.05%) (Figs. 4A, B, C, and D). The mean lesion volume (units of 0. 1 mm3) at the last time point before death was significantly different among the four patterns ( P = 0.03, Tukey – Kramer’s t test). The largest mean volume was seen in fluctuating lesions (49.81 F 40.78), followed by expanding lesions (35.96 F 27.5),
Lesion type
Number Percentage Mean volume of of lesions at last time lesions (%) point (0.1 mm3)
Expanding Expanding – retracting Fluctuating Stable
79 34
48.50 20.85
36.0* (F 27.5) 54.1 21.8* (F 15.3) 15.0
27 23
16.60 14.05
49.8* (F 40.8) 28.8 6.2* (F 2.5) 2.1
* P = 0.03.
Percentage of total volume at last time point ‘‘Contribution to morbidity’’ (%)
580
I. Pirko et al. / NeuroImage 21 (2004) 576–582
Fig. 4. (A) Expanding lesions. (B) Expanding – retracting lesions. (C) Fluctuating lesions. (D) Stable lesions. Four distinct chronological patterns of lesion development were found in the studied TMEV-infected IFN-gR / mice. The curves illustrate each of these patterns. Every studied lesion is shown with marks on these graphs; however, only a few selected lesions are represented with connected lines to illustrate the individual chronological pattern.
opment. Expanding (n = 79, enhancing = 67) and expanding – retracting lesions (n = 34, enhancing = 28) were associated with enhancement most of the time (85.3% and 83.5%, respectively). However, only 56.5% of stable lesions was associated with gadolinium enhancement (n = 23, enhancing = 13). Furthermore, lesions that appeared 43 days or later after the infection did not show enhancement in more than 92% of the cases. Altogether, 28 lesions appeared after day 43, yet only two showed gadolinium enhancement. Those two were expanding lesions. The other 26 lesions were either stable (10), expanding – retracting lesions (9), or expanding (7). In addition, gadolinium enhancement was relatively short-lived. Enhancement never persisted for more than 7 days.
Discussion The aim of this study was to determine the MRI characteristics of TMEV infection in a mouse model characterized by rapid and
steady progression of demyelination without periods of remission. In the TMEV model, we found no lesions that resolved. However, in murine EAE, some gadolinium-enhancing lesions do not become permanently visible on subsequent T2-weighted scans (Jordan et al., 1999). It is possible that such phenomena occurred in our model as well. This may have been detected in the TMEV model with even higher gadolinium doses, or with the application of a magnetization transfer pulse to increase the detectability of enhancing areas (Bastianello et al., 1998). Also, the fact that no lesions resolved in the TMEV model may have been related to a ‘‘threshold effect’’, because the lesions that eventually resolve in EAE are usually small and may elude detection (Jordan et al., 1999). On high field strength (above 4 T) small-animal MRI scans, there may be very small areas of high T1 signal even in the absence of gadolinium injection (Hart et al., 1998). These areas tend to be adjacent to other lesions, and thus far, no clear pathological correlates have been found as to what they may represent. We also noticed the presence of similar areas that rendered detection of
I. Pirko et al. / NeuroImage 21 (2004) 576–582
581
Fig. 5. Total lesion load of all studied lesions—data are stratified by lesion type. This figure illustrates the total volume of all studied lesions. The figure also shows the breakdown of different chronological lesion types to illustrate the unequal contribution of lesion types to the total volume.
similarly small, but truly enhancing lesions more difficult. In human MS, not all gadolinium-enhancing ‘‘early inflammatory’’ lesions will become permanently visible on subsequent T2-weighted scans (Barkhof and van Walderveen, 1999; Ciccarelli et al., 1999; Rovaris et al., 1999). Up to 8% of all lesions may disappear both from the T1 and the T2 image by 5 months (Ciccarelli et al., 1999). We found that the formation of T2-weighted hyperintense lesions followed four distinct patterns. Some lesions showed continuous or fluctuating growth, others showed growth followed by retraction, but not full resolution, yet others remained stable. Such behavior of T2-weighted lesions has not been reported in animal models before. Furthermore, this novel finding has also not been reported in human MS to our knowledge. We would also like to emphasize that these four patterns are not histological patterns, but chronological patterns as observed by MRI. At a given time point, even on MRI, it cannot be determined whether a lesion belongs to one of these categories or the other. It is possible that there are underlying histological differences among these lesions at certain time points, and we plan to study this further. We also noticed a dramatic increase in total lesion load over the last week of the illness. We could not find any similar reports in the literature neither in animal models nor in human MS. As shown in Fig. 2, the number of lesions increased linearly in this model, whereas the total lesion load clearly showed a biphasic distribution (Fig. 3). Furthermore, lesions that appeared in the last week of illness were mostly stable lesions, which were very small. Therefore, the expansive increase in lesion load was not explained merely by new lesion formation, but was in a large part related to the growth of existing lesions. Also of interest was that most lesions were in the brainstem, which may explain the uniformly morbid outcome of the illness (see movie files).
Another surprising finding was the lack of observable gadolinium enhancement in some of the late-appearing lesions. The number of lesions increased linearly; thus, new lesions were still seen even in the last few days of the illness. It has been reported in EAE that newly appearing small lesions adjacent to preexisting lesions (especially to relatively new lesions) may not show gadolinium enhancement on high-field scans (Hart et al., 1998). In the EAE model, most of the gadolinium-enhancing lesions in the early inflammatory stage were in periventricular or juxtacortical areas, whereas lesions located elsewhere were not detected as enhancing lesions before their appearance on T2 scans. However, their detection may be increased by the use of a magnetization transfer pulse (Jordan et al., 1999). The term ‘‘nonenhancing active lesions’’ has been coined to describe this phenomenon (Hart et al., 1998). Others have postulated that in these lesions, there actually is blood – brain barrier disruption. However, the constant leakage of large albumin molecules may not permit the leakage of small gadolinium – chelator complexes. Furthermore, whereas albumin is present continuously in the bloodstream, the gadolinium-based contrast material is only present temporarily and has a very rapid clearance, even when used in triple dose (Hart et al., 1998; Xu et al., 1998). Theoretically, by application of even higher or repeated doses of gadolinium, such leakage may be visualized. In addition, by the use of paramagnetically labeled albumin, the same goal could be achieved. We plan to study this phenomenon further with labeled albumin. It is also possible that new lesions may form without significant disruption of the blood – brain barrier, especially in areas of locally high lesion load, by nonconcentric expansion of enlarging, but no longer enhancing lesions. Another theoretical possibility is that the enhancement is very short-lived, and despite the frequent scan interval, we simply missed the early inflammatory stages. It is
582
I. Pirko et al. / NeuroImage 21 (2004) 576–582
known that in EAE models, gadolinium enhancement is a short phenomenon in general. Enhancement is usually observed only for a few days, up to a week (Hart et al., 1998; Jordan et al., 1999). In conclusion, this TMEV-induced demyelinating model of MS shows several novel MRI features that may aid in our understanding of acute progressive demyelinating diseases. Our team plans to further investigate these findings with the inclusion of histopathology to MRI studies. These insights may help evaluate strategies directed at treating acute progressive forms of MS. References Barkhof, F., van Walderveen, M., 1999. Characterization of tissue damage in multiple sclerosis by nuclear magnetic resonance. Philos. Trans. R. Soc. Lond., B Biol. Sci. 354, 1675 – 1686. Bastianello, S., Gasperini, C., Paolillo, A., Giugni, E., Ciccarelli, O., Sormani, M.P., Horsfield, M.A., Rovaris, M., Pozzilli, C., Filippi, M., 1998. Sensitivity of enhanced MR in multiple sclerosis: effects of contrast dose and magnetization transfer contrast. AJNR Am. J. Neuroradiol. 19, 1863 – 1867. Brahic, M., 2002. Theiler’s virus infection of the mouse, or: of the importance of studying animal models. Virology 301, 1 – 5. Brahic, M., Bureau, J.F., 1998. Genetics of susceptibility to Theiler’s virus infection. BioEssays 20, 627 – 633. Ciccarelli, O., Giugni, E., Paolillo, A., Mainero, C., Gasperini, C., Bastianello, S., Pozzilli, C., 1999. Magnetic resonance outcome of new enhancing lesions in patients with relapsing – remitting multiple sclerosis. Eur. J. Neurol. 6, 455 – 459. Dal Canto, M.C., Melvold, R.W., Kim, B.S., Miller, S.D., 1995. Two models of multiple sclerosis: experimental allergic encephalomyelitis (EAE) and Theiler’s murine encephalomyelitis virus (TMEV) infection. A pathological and immunological comparison. Microsc. Res. Tech. 32, 215 – 229. Hart, B.A., Bauer, J., Muller, H.J., Melchers, B., Nicolay, K., Brok, H., Bontrop, R.E., Lassmann, H., Massacesi, L., 1998. Histopathological characterization of magnetic resonance imaging-detectable brain white matter lesions in a primate model of multiple sclerosis: a correlative study in the experimental autoimmune encephalomyelitis model in common marmosets (Callithrix jacchus). Am. J. Pathol. 153, 649 – 663. Johnson, A.J., Upshaw, J., Pavelko, K.D., Rodriguez, M., Pease, L.R., 2001. Preservation of motor function by inhibition of CD8+ virus peptidespecific T cells in Theiler’s virus infection. FASEB J. 15, 2760 – 2762.
Jordan, E.K., McFarland, H.I., Lewis, B.K., Tresser, N., Gates, M.A., Johnson, M., Lenardo, M., Matis, L.A., McFarland, H.F., Frank, J.A., 1999. Serial MR imaging of experimental autoimmune encephalomyelitis induced by human white matter or by chimeric myelin-basic and proteolipid protein in the common marmoset. AJNR Am. J. Neuroradiol. 20, 965 – 976. McDonald, W.I., Compston, A., Edan, G., Goodkin, D., Hartung, H.P., Lublin, F.D., McFarland, H.F., Paty, D.W., Polman, C.H., Reingold, S.C., et al., 2001. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann. Neurol. 50, 121 – 127. Murray, P.D., McGavern, D.B., Pease, L.R., Rodriguez, M., 2002. Cellular sources and targets of IFN-gamma-mediated protection against viral demyelination and neurological deficits. Eur. J. Immunol. 32, 606 – 615. Noseworthy, J.H., Lucchinetti, C., Rodriguez, M., Weinshenker, B.G., 2000. Multiple sclerosis. N. Engl. J. Med. 343, 938 – 952. Rodriguez, M., Pavelko, K., Coffman, R.L., 1995. Gamma interferon is critical for resistance to Theiler’s virus-induced demyelination. J. Virol. 69, 7286 – 7290. Rovaris, M., Mastronardo, G., Prandini, F., Bastianello, S., Comi, G., Filippi, M., 1999. Short-term evolution of new multiple sclerosis lesions enhancing on standard and triple dose gadolinium-enhanced brain MRI scans. J. Neurol. Sci. 164, 148 – 152. Thorpe, J.W., Barker, G.J., MacManus, D.G., Moseley, I.F., Tofts, P.S., Miller, D.H., 1994. Detection of multiple sclerosis by magnetic resonance imaging. Lancet 344, 1235. van Waesberghe, J.H., van Walderveen, M.A., Castelijns, J.A., Scheltens, P., Lycklama a Nijeholt, G.J., Polman, C.H., Barkhof, F., 1998. Patterns of lesion development in multiple sclerosis: longitudinal observations with T1-weighted spin-echo and magnetization transfer MR. AJNR Am. J. Neuroradiol. 19, 675 – 683. van Walderveen, M.A., Barkhof, F., Tas, M.W., Polman, C., Frequin, S.T., Hommes, O.R., Thompson, A.J., Valk, J., 1998. Patterns of brain magnetic resonance abnormalities on T2-weighted spin echo images in clinical subgroups of multiple sclerosis: a large cross-sectional study. Eur. Neurol. 40, 91 – 98. Xu, S., Jordan, E.K., Brocke, S., Bulte, J.W., Quigley, L., Tresser, N., Ostuni, J.L., Yang, Y., McFarland, H.F., Frank, J.A., 1998. Study of relapsing remitting experimental allergic encephalomyelitis SJL mouse model using MION-46L enhanced in vivo MRI: early histopathological correlation. J. Neurosci. Res. 52, 549 – 558. Young, I.R., Hall, A.S., Pallis, C.A., Legg, N.J., Bydder, G.M., Steiner, R.E., 1981. Nuclear magnetic resonance imaging of the brain in multiple sclerosis. Lancet 2, 1063 – 1066.
Dynamics of MRI lesion development in an animal model of viral-induced acute progressive CNS demyelination Istvan Pirko, a Jeff Gamez, b Aaron J. Johnson, b Slobodan I. Macura, c,d and Moses Rodriguez a,b,* a
Department of Neurology, Mayo Clinic, Rochester, MN, USA Department of Immunology, Mayo Clinic, Rochester, MN, USA c Department of Biochemistry, Mayo Clinic, Rochester, MN, USA d NMR Core Facility, Mayo Clinic, Rochester, MN, USA b
Received 21 May 2003; revised 4 August 2003; accepted 15 September 2003
Theiler’s murine encephalitis virus (TMEV) infection in mice is an established model of CNS demyelinating diseases. The aim of the study was to determine the chronological pattern of lesion development in this model of monophasic fulminant demyelinating disease. We followed six highly susceptible interferon-gamma receptor knockout mice with serial in vivo brain magnetic resonance imaging (MRI) studies to determine changes in overall T2 lesion load and gadolinium enhancement. Altogether, 163 individual lesions were followed over 52 days. The number of lesions increased linearly with time. Four chronological patterns of lesion development were seen: (a) expanding lesions (48.5% of all lesions, 54.05% volume contribution); (b) expanding – retracting lesions (20.85% of all lesions, 15.03% volume contribution); (c) fluctuating lesions (16.6% of all lesions, 28.8% volume contribution); (d) stable lesions (14.05% of all lesions, 2.12% volume contribution). Gadolinium enhancement was not seen in the evolution of every lesion. Enhancement was both time- and lesion type-dependent. Early in the disease course ( < 43 days after infection), enhancement was almost always seen, later on (>43 days after infection) it was only seen in 8% of new lesions. All of fluctuating, 85.3% of expanding, 83.5% of expanding – retracting, and 56.5% of stable lesions were associated with gadolinium enhancement. We conclude that the MRI features of TMEV-induced demyelination in this model showed four unique chronological patterns, and inconsistent gadolinium enhancement. These novel findings may provide new insights into the pathogenesis of acute fulminant multiple sclerosis (MS). D 2003 Elsevier Inc. All rights reserved. Keywords: Lesion development; CNS demyelination; MRI
Introduction Multiple sclerosis (MS) is the most common inflammatory demyelinating disease of the central nervous system (Noseworthy et al., 2000). The disease has important socioeconomic impact in that
* Corresponding author. Department of Neurology, Mayo Clinic, Rochester, MN 55905. Fax: +1-507-284-1086. E-mail address: [email protected] (M. Rodriguez). Available online on ScienceDirect (www.sciencedirect.com.) 1053-8119/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2003.09.037
it is the leading cause of disability in young adults in developed countries. Magnetic resonance imaging (MRI) is a well-established tool in diagnosing and monitoring MS patients (Thorpe et al., 1994; Young et al., 1981). In addition, it has also become a frequently used secondary outcome measure in many MS-related therapeutic clinical trials (Barkhof and van Walderveen, 1999). The role of MRI in MS has increased given the newly established McDonald’s criteria, which puts greater emphasis on MRI studies (McDonald et al., 2001). The dynamics of lesion formation in MS has been studied (van Waesberghe et al., 1998; van Walderveen et al., 1998). T2-weighted sequences are especially sensitive to structural alterations of CNS tissue. All stages of MS lesion formation, from early inflammatory to late chronic, are visualized as hyperintense regions on T2weighted scans. Active lesions can be distinguished from chronic lesions by their increased permeability of the blood – brain barrier, which is demonstrated as enhancement of the gadolinium contrast on T1-weighted scans. With the advent of in vivo microscopic resolution small animal MRI, similar monitoring has become available for frequently used animal models of MS, including Theiler’s murine Encephalitis Virus (TMEV) infection. TMEV (Dal Canto et al., 1995) is a picornavirus that normally causes gastrointestinal infection in mice, occasionally accompanied by encephalitis. However, if the virus is inoculated intracerebrally, then a biphasic disease develops in susceptible strains of mice. The first phase lasts approximately 2 weeks. This phase is characterized by encephalitis and meningitis. In the second phase (beginning within 3 weeks after infection), immune-mediated demyelination is observed, which is similar to human demyelinating disorders (Brahic, 2002; Brahic and Bureau, 1998). Even though this model has been extensively studied since 1934 (Brahic and Bureau, 1998), its MRI features have not been characterized. The goal of our study was to characterize MRI lesion development in the TMEV model of MS in interferon-gamma receptor knockout mice (IFN-gR / ). We chose to do our MRI investigations in these mice because they develop an acute progressive demyelinating illness that results in death within 6 – 8 weeks, due to extensive brain and brainstem lesions (Johnson et al., 2001; Murray et al., 2002; Rodriguez et al., 1995). This is in contrast to
I. Pirko et al. / NeuroImage 21 (2004) 576–582
the disease that develops in the most commonly studied SJL/J strain. SJL/J mice develop a chronic demyelinating illness that is almost exclusively limited to the spinal cord, with little or no brain pathology observed. We studied in IFN-gR / mice the significance of gadolinium enhancement and the dynamics of T2 lesion formation by three-dimensional microscopic resolution in vivo magnetic resonance imaging at eight time points during the disease course.
577
the right parietal cortex. Intracerebral injection results in greater than 98% incidence of demyelination in mice of susceptible haplotype with only a rare fatality following injection. All animals develop mild viral encephalitis, which resolves within 14 days after injection. The animals then develop a chronic demyelinating disease in the brain and spinal cord that progresses over months. Magnetic resonance imaging (MRI)
Materials and methods Mice 129-Ifngrtm1 (IFN-gR / ) mice were generously provided to us by Michel Aguet (Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland). Animals were housed and bred in Mayo’s animal care facility. All animal studies conformed to the guidelines for use of animals by the Mayo Clinic and the National Institutes of Health. Theiler’s virus infection TMEV-induced demyelination was produced by intracerebral virus injection of 4 – 8-week-old mice anesthetized with methohexital. With a 27-gauge needle attached to a Hamilton syringe, a 10-Al volume containing 200,000 plaque-forming units (PFU) purified Daniel’s strain of TMEV was injected intracerebrally in
Six IFN-gR / mice were followed with high resolution 3D MRI scans on days 4, 10, 20, 27, 35, 43, 47, and 52 after TMEV infection. The choice of time points was in part determined by scanner availability. One animal expired on day 35. All five remaining animals expired on or after day 52 but before the planned subsequent imaging time point. The total scan time for all experiments was 136 h. The MRI examinations were performed in a Bruker Avance 300 MHz (7 T) vertical bore NMR spectrometer equipped with ‘‘miniimaging’’ accessories. The 37jC core temperature of the animals was maintained by a thermocouple-based system. Inhalational isofluran anesthesia (1.5% in oxygen) was delivered via a nose cone during the imaging procedure. Following the i.v. injection of ‘‘triple-dose’’ Magnevist (0.2 ml/100 g body weight of 1:3 physiologic saline-diluted Magnevist solution) T1-weighted volume-acquisition spin-echo sequences were used to visualize gadolinium enhancing lesions (TR: 400 ms, TE: 6.2 ms, FOV: 4 2.5 2.5 cm, matrix: 256 160 160). T2-
Fig. 1. Example of lesion development in an IFN-gR / mouse infected with TMEV. (A) Day 4, no lesions. (B) Day 10, four lesions; (C) Day 20, eight lesions; (D) Day 27, 14 lesions; (E) Day 35, 21 lesions; (F) Day 43, 29 lesions; (G) Day 47, 34 lesions; (H) Day 52, 36 lesions. The images represent 3D volume rendering resampled from T2-weighted 3D datasets of the studied animals. Lesions were extracted as detailed in Materials and methods, and object maps were generated to describe the lesions. Every lesion was color coded individually to help identification.
578
I. Pirko et al. / NeuroImage 21 (2004) 576–582
weighted imaging was performed using a 3D RARE volume acquisition sequence (TR: 1500 ms, TE: 70 ms, Flipback: ON, RARE factor: 16, FOV: 3.5 3.5 3.5 cm, matrix: 160 160 160). Analysis of 3D image sets: brain extraction, coregistration, segmentation, and volumetry Analyze 4.0 and 5.0, developed by the Biomedical Imaging Resource at the Mayo Clinic, was used for processing and analysis of the obtained 3D image sets. The datasets were segmented, and the brains were extracted using the Object Extractor and Image Edit tools. To correct for potential misalignment of the extracted brains at different time points, a 3D surface matching algorithm was used. This algorithm generates a rotation matrix. The rotated and coregistered 3D brain images were resampled with windowed sync interpolation of the original images. The co-registered images were then segmented using the 3D ROI Analysis Tool. A semiautomated intensity-based seed-growing algorithm was used to generate object maps. The object maps defined sub-volumes of the brain. Each subvolume is a stand-alone, numbered, individually identified, threedimensional brain lesion. The generated object maps were then applied to the subsequent time point, and were corrected as appropriate with the same seed-growing algorithm so that changes
in lesion number, volume, and shape could be accounted for. Because we used semiautomated tools, there was a human factor involved in this analysis. To minimize this, the involved investigators were trained for several weeks on test datasets, and their intra- and inter-rater reliability was repeatedly tested and was found to be superior (>95%). After the generation of object maps describing each time point in every animal, the 3D ROI Scan Tool was used to calculate individual and total lesion volumes.
Results MRI lesion development as a function of time Altogether, the animals developed 163 lesions. The number of lesions increased linearly in every animal (Figs. 1 and 2). There was no significant difference between the numbers of lesions among the animals at the studied time points. The total lesion volume per animal increased in a biphasic fashion. There was an initial linear increase until day 43; afterwards, a more rapid expansion of total lesion load was seen (Fig. 3). No significant difference was observed among lesion loads of animals at the last time point before death.
Fig. 2. Number of lesions per animal. The number of lesions increased linearly in all studied animals. The figure shows data of all animals studied. The mean values are also presented.
I. Pirko et al. / NeuroImage 21 (2004) 576–582
579
Fig. 3. Changes in lesion load (total lesion volume per time point) for each animal. At every represented time point, the total volume of all observed lesions was determined for each animal. The curves showed a biphasic growth of total lesion load with an initial linear phase, which was followed by a phase of rapid increase after day 42. The mean lesion volumes are also presented.
All 163 lesions were followed individually. None of the lesions observed on T2-weighted scans or gadolinium-enhanced T1weighted scans resolved during the illness. MRI datasets visualized on 3D movies To visually demonstrate the development of lesions in this model, 3D movies were generated using MRI datasets obtained from one mouse. Movie 1 shows a 3D rotation of a mouse brain extracted from the in vivo dataset. Every full rotation represents a new time point. The brain is rendered as a semitransparent object to allow for visualization of the individually color-coded lesions. Movie 2 is a time-lapse movie: the same semitransparent brain is seen from above; the developing lesions are color coded individually with the same method.
and expanding – retracting lesions (21.76 F 15.27). The mean volume of stable lesions was the smallest (6.17 F 2.51). Expanding lesions represented 54.05% of the total lesion load at the last time point before morbidity, followed by fluctuating (28.8%) and expanding – retracting lesions (15.03%). The total volume of stable lesions contributed least to the total lesion load, and altogether represented only 2.12% of the combined total lesion load of all animals at the last time point (Fig. 5). Gadolinium enhancement was not an absolute prerequisite for lesion formation Gadolinium enhancement was not seen in a consistent fashion during the formation of new lesions. Fluctuating lesions (n = 27) always showed this phenomenon in the first 3 – 7 days of devel-
Four chronological patterns of lesion development were observed
Table 1 Main characteristics of lesion types
The lesion formation as observed on T2-weighted scans followed one of four patterns (Table 1). From most common to least common, these were (a) expanding (continuous growth, n = 79, 48.5% of all lesions), (b) expanding – retracting (growth followed by partial resolution, n = 34, 20.85%), (c) fluctuating (growth – retraction – growth, n = 27, 16.6%), or (d) stable lesions (n = 23, 14.05%) (Figs. 4A, B, C, and D). The mean lesion volume (units of 0. 1 mm3) at the last time point before death was significantly different among the four patterns ( P = 0.03, Tukey – Kramer’s t test). The largest mean volume was seen in fluctuating lesions (49.81 F 40.78), followed by expanding lesions (35.96 F 27.5),
Lesion type
Number Percentage Mean volume of of lesions at last time lesions (%) point (0.1 mm3)
Expanding Expanding – retracting Fluctuating Stable
79 34
48.50 20.85
36.0* (F 27.5) 54.1 21.8* (F 15.3) 15.0
27 23
16.60 14.05
49.8* (F 40.8) 28.8 6.2* (F 2.5) 2.1
* P = 0.03.
Percentage of total volume at last time point ‘‘Contribution to morbidity’’ (%)
580
I. Pirko et al. / NeuroImage 21 (2004) 576–582
Fig. 4. (A) Expanding lesions. (B) Expanding – retracting lesions. (C) Fluctuating lesions. (D) Stable lesions. Four distinct chronological patterns of lesion development were found in the studied TMEV-infected IFN-gR / mice. The curves illustrate each of these patterns. Every studied lesion is shown with marks on these graphs; however, only a few selected lesions are represented with connected lines to illustrate the individual chronological pattern.
opment. Expanding (n = 79, enhancing = 67) and expanding – retracting lesions (n = 34, enhancing = 28) were associated with enhancement most of the time (85.3% and 83.5%, respectively). However, only 56.5% of stable lesions was associated with gadolinium enhancement (n = 23, enhancing = 13). Furthermore, lesions that appeared 43 days or later after the infection did not show enhancement in more than 92% of the cases. Altogether, 28 lesions appeared after day 43, yet only two showed gadolinium enhancement. Those two were expanding lesions. The other 26 lesions were either stable (10), expanding – retracting lesions (9), or expanding (7). In addition, gadolinium enhancement was relatively short-lived. Enhancement never persisted for more than 7 days.
Discussion The aim of this study was to determine the MRI characteristics of TMEV infection in a mouse model characterized by rapid and
steady progression of demyelination without periods of remission. In the TMEV model, we found no lesions that resolved. However, in murine EAE, some gadolinium-enhancing lesions do not become permanently visible on subsequent T2-weighted scans (Jordan et al., 1999). It is possible that such phenomena occurred in our model as well. This may have been detected in the TMEV model with even higher gadolinium doses, or with the application of a magnetization transfer pulse to increase the detectability of enhancing areas (Bastianello et al., 1998). Also, the fact that no lesions resolved in the TMEV model may have been related to a ‘‘threshold effect’’, because the lesions that eventually resolve in EAE are usually small and may elude detection (Jordan et al., 1999). On high field strength (above 4 T) small-animal MRI scans, there may be very small areas of high T1 signal even in the absence of gadolinium injection (Hart et al., 1998). These areas tend to be adjacent to other lesions, and thus far, no clear pathological correlates have been found as to what they may represent. We also noticed the presence of similar areas that rendered detection of
I. Pirko et al. / NeuroImage 21 (2004) 576–582
581
Fig. 5. Total lesion load of all studied lesions—data are stratified by lesion type. This figure illustrates the total volume of all studied lesions. The figure also shows the breakdown of different chronological lesion types to illustrate the unequal contribution of lesion types to the total volume.
similarly small, but truly enhancing lesions more difficult. In human MS, not all gadolinium-enhancing ‘‘early inflammatory’’ lesions will become permanently visible on subsequent T2-weighted scans (Barkhof and van Walderveen, 1999; Ciccarelli et al., 1999; Rovaris et al., 1999). Up to 8% of all lesions may disappear both from the T1 and the T2 image by 5 months (Ciccarelli et al., 1999). We found that the formation of T2-weighted hyperintense lesions followed four distinct patterns. Some lesions showed continuous or fluctuating growth, others showed growth followed by retraction, but not full resolution, yet others remained stable. Such behavior of T2-weighted lesions has not been reported in animal models before. Furthermore, this novel finding has also not been reported in human MS to our knowledge. We would also like to emphasize that these four patterns are not histological patterns, but chronological patterns as observed by MRI. At a given time point, even on MRI, it cannot be determined whether a lesion belongs to one of these categories or the other. It is possible that there are underlying histological differences among these lesions at certain time points, and we plan to study this further. We also noticed a dramatic increase in total lesion load over the last week of the illness. We could not find any similar reports in the literature neither in animal models nor in human MS. As shown in Fig. 2, the number of lesions increased linearly in this model, whereas the total lesion load clearly showed a biphasic distribution (Fig. 3). Furthermore, lesions that appeared in the last week of illness were mostly stable lesions, which were very small. Therefore, the expansive increase in lesion load was not explained merely by new lesion formation, but was in a large part related to the growth of existing lesions. Also of interest was that most lesions were in the brainstem, which may explain the uniformly morbid outcome of the illness (see movie files).
Another surprising finding was the lack of observable gadolinium enhancement in some of the late-appearing lesions. The number of lesions increased linearly; thus, new lesions were still seen even in the last few days of the illness. It has been reported in EAE that newly appearing small lesions adjacent to preexisting lesions (especially to relatively new lesions) may not show gadolinium enhancement on high-field scans (Hart et al., 1998). In the EAE model, most of the gadolinium-enhancing lesions in the early inflammatory stage were in periventricular or juxtacortical areas, whereas lesions located elsewhere were not detected as enhancing lesions before their appearance on T2 scans. However, their detection may be increased by the use of a magnetization transfer pulse (Jordan et al., 1999). The term ‘‘nonenhancing active lesions’’ has been coined to describe this phenomenon (Hart et al., 1998). Others have postulated that in these lesions, there actually is blood – brain barrier disruption. However, the constant leakage of large albumin molecules may not permit the leakage of small gadolinium – chelator complexes. Furthermore, whereas albumin is present continuously in the bloodstream, the gadolinium-based contrast material is only present temporarily and has a very rapid clearance, even when used in triple dose (Hart et al., 1998; Xu et al., 1998). Theoretically, by application of even higher or repeated doses of gadolinium, such leakage may be visualized. In addition, by the use of paramagnetically labeled albumin, the same goal could be achieved. We plan to study this phenomenon further with labeled albumin. It is also possible that new lesions may form without significant disruption of the blood – brain barrier, especially in areas of locally high lesion load, by nonconcentric expansion of enlarging, but no longer enhancing lesions. Another theoretical possibility is that the enhancement is very short-lived, and despite the frequent scan interval, we simply missed the early inflammatory stages. It is
582
I. Pirko et al. / NeuroImage 21 (2004) 576–582
known that in EAE models, gadolinium enhancement is a short phenomenon in general. Enhancement is usually observed only for a few days, up to a week (Hart et al., 1998; Jordan et al., 1999). In conclusion, this TMEV-induced demyelinating model of MS shows several novel MRI features that may aid in our understanding of acute progressive demyelinating diseases. Our team plans to further investigate these findings with the inclusion of histopathology to MRI studies. These insights may help evaluate strategies directed at treating acute progressive forms of MS. References Barkhof, F., van Walderveen, M., 1999. Characterization of tissue damage in multiple sclerosis by nuclear magnetic resonance. Philos. Trans. R. Soc. Lond., B Biol. Sci. 354, 1675 – 1686. Bastianello, S., Gasperini, C., Paolillo, A., Giugni, E., Ciccarelli, O., Sormani, M.P., Horsfield, M.A., Rovaris, M., Pozzilli, C., Filippi, M., 1998. Sensitivity of enhanced MR in multiple sclerosis: effects of contrast dose and magnetization transfer contrast. AJNR Am. J. Neuroradiol. 19, 1863 – 1867. Brahic, M., 2002. Theiler’s virus infection of the mouse, or: of the importance of studying animal models. Virology 301, 1 – 5. Brahic, M., Bureau, J.F., 1998. Genetics of susceptibility to Theiler’s virus infection. BioEssays 20, 627 – 633. Ciccarelli, O., Giugni, E., Paolillo, A., Mainero, C., Gasperini, C., Bastianello, S., Pozzilli, C., 1999. Magnetic resonance outcome of new enhancing lesions in patients with relapsing – remitting multiple sclerosis. Eur. J. Neurol. 6, 455 – 459. Dal Canto, M.C., Melvold, R.W., Kim, B.S., Miller, S.D., 1995. Two models of multiple sclerosis: experimental allergic encephalomyelitis (EAE) and Theiler’s murine encephalomyelitis virus (TMEV) infection. A pathological and immunological comparison. Microsc. Res. Tech. 32, 215 – 229. Hart, B.A., Bauer, J., Muller, H.J., Melchers, B., Nicolay, K., Brok, H., Bontrop, R.E., Lassmann, H., Massacesi, L., 1998. Histopathological characterization of magnetic resonance imaging-detectable brain white matter lesions in a primate model of multiple sclerosis: a correlative study in the experimental autoimmune encephalomyelitis model in common marmosets (Callithrix jacchus). Am. J. Pathol. 153, 649 – 663. Johnson, A.J., Upshaw, J., Pavelko, K.D., Rodriguez, M., Pease, L.R., 2001. Preservation of motor function by inhibition of CD8+ virus peptidespecific T cells in Theiler’s virus infection. FASEB J. 15, 2760 – 2762.
Jordan, E.K., McFarland, H.I., Lewis, B.K., Tresser, N., Gates, M.A., Johnson, M., Lenardo, M., Matis, L.A., McFarland, H.F., Frank, J.A., 1999. Serial MR imaging of experimental autoimmune encephalomyelitis induced by human white matter or by chimeric myelin-basic and proteolipid protein in the common marmoset. AJNR Am. J. Neuroradiol. 20, 965 – 976. McDonald, W.I., Compston, A., Edan, G., Goodkin, D., Hartung, H.P., Lublin, F.D., McFarland, H.F., Paty, D.W., Polman, C.H., Reingold, S.C., et al., 2001. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann. Neurol. 50, 121 – 127. Murray, P.D., McGavern, D.B., Pease, L.R., Rodriguez, M., 2002. Cellular sources and targets of IFN-gamma-mediated protection against viral demyelination and neurological deficits. Eur. J. Immunol. 32, 606 – 615. Noseworthy, J.H., Lucchinetti, C., Rodriguez, M., Weinshenker, B.G., 2000. Multiple sclerosis. N. Engl. J. Med. 343, 938 – 952. Rodriguez, M., Pavelko, K., Coffman, R.L., 1995. Gamma interferon is critical for resistance to Theiler’s virus-induced demyelination. J. Virol. 69, 7286 – 7290. Rovaris, M., Mastronardo, G., Prandini, F., Bastianello, S., Comi, G., Filippi, M., 1999. Short-term evolution of new multiple sclerosis lesions enhancing on standard and triple dose gadolinium-enhanced brain MRI scans. J. Neurol. Sci. 164, 148 – 152. Thorpe, J.W., Barker, G.J., MacManus, D.G., Moseley, I.F., Tofts, P.S., Miller, D.H., 1994. Detection of multiple sclerosis by magnetic resonance imaging. Lancet 344, 1235. van Waesberghe, J.H., van Walderveen, M.A., Castelijns, J.A., Scheltens, P., Lycklama a Nijeholt, G.J., Polman, C.H., Barkhof, F., 1998. Patterns of lesion development in multiple sclerosis: longitudinal observations with T1-weighted spin-echo and magnetization transfer MR. AJNR Am. J. Neuroradiol. 19, 675 – 683. van Walderveen, M.A., Barkhof, F., Tas, M.W., Polman, C., Frequin, S.T., Hommes, O.R., Thompson, A.J., Valk, J., 1998. Patterns of brain magnetic resonance abnormalities on T2-weighted spin echo images in clinical subgroups of multiple sclerosis: a large cross-sectional study. Eur. Neurol. 40, 91 – 98. Xu, S., Jordan, E.K., Brocke, S., Bulte, J.W., Quigley, L., Tresser, N., Ostuni, J.L., Yang, Y., McFarland, H.F., Frank, J.A., 1998. Study of relapsing remitting experimental allergic encephalomyelitis SJL mouse model using MION-46L enhanced in vivo MRI: early histopathological correlation. J. Neurosci. Res. 52, 549 – 558. Young, I.R., Hall, A.S., Pallis, C.A., Legg, N.J., Bydder, G.M., Steiner, R.E., 1981. Nuclear magnetic resonance imaging of the brain in multiple sclerosis. Lancet 2, 1063 – 1066.