Volumes of brain atrophy and plaques correlated with neurological disability in secondary progressive multiple sclerosis

Journal of the Neurological Sciences 165 (1999) 36–42

Volumes of brain atrophy and plaques correlated with neurological disability in secondary progressive multiple sclerosis P. Dastidar

a,d ,

*, T. Heinonen b , T. Lehtimaki ¨ c,d , M. Ukkonen e , J. Peltola e , T. Erila¨ e , E. Laasonen a,d , I. Elovaara e a

Department of Diagnostic Radiology, Tampere University Hospital, Tampere, Finland b Ragnar Granit Institute, University of Technology, Tampere, Finland c Laboratory for Atherosclerosis Genetics, Department of Clinical Chemistry, Tampere University Hospital, Tampere, Finland d Medical School, Tampere University, Tampere, Finland e Department of Neurology, Tampere University Hospital, Tampere, Finland Received 31 July 1998; received in revised form 3 March 1999; accepted 5 March 1999

Abstract The objectives of the present study was to correlate the segmented magnetic resonance imaging (MRI) volumes of intracranial cerebrospinal fluid (CSF) spaces (expressing the extent of brain atrophy) and cerebral plaques with the neurological disability in secondary progressive multiple sclerosis (MS). Earlier studies have mainly correlated MS plaques and neurological disability measured by expanded disability status scale (EDSS). The data on the association between brain atrophy and EDSS or regional functional scoring scale (RFSS) are very limited. We measured the volumes of intracranial CSF spaces in 28 patients with secondary progressive MS using MRI, and semiautomatic segmentation software. The volumes of T1-weighted hypointense and T2-weighted hyperintense MS plaques were also measured. In multiple regression analysis, increasing volumes of total (P50.006) and relative (P50.005) intracranial CSF spaces were significantly associated with worsening neurological disability as expressed by EDSS. No associations were found between these intracranial CSF space volumes and total RFSS scores. The mean volume of T2-weighted plaques showed a tendency to associate with total RFSS score (r50.40, P50.03), but no correlations were detected between T1- or T2-weighted plaque volumes and EDSS. The application of a new segmentation technique in quantifying intracranial cerebrospinal fluid spaces allowed an exact and sensitive way of assessing brain atrophy. The associations between brain atrophy and neurological disability expressed by EDSS suggests that the effect of MS therapies should be evaluated by measurement of brain atrophy.  1999 Elsevier Science B.V. All rights reserved. Keywords: Multiple sclerosis; EDSS; RFSS; Magnetic resonance; Image processing; Magnetic resonance; Volume measurement

1. Introduction Magnetic resonance imaging (MRI) with its new developments is the most important imaging technique today in the diagnosis of multiple sclerosis (MS) [18,36,42]. A more reliable and reproducible analysis of MRI lesions can be now obtained with new computer-assisted quantitative *Corresponding author. Correspondence address: Tampere University Hospital, Department of Diagnostic Radiology, P.O. Box 2000, FIN33521 Tampere, Finland. Tel.: 1358-3-247-6504; fax: 1358-3-247-5586. 0022-510X / 99 / $ – see front matter PII: S0022-510X( 99 )00071-4

MRI techniques [41]. The process of detecting and delineating objects in images referred to as image segmentation embodies a vast amount of literature [7]. The use of various segmentation techniques has become an important part in the accurate quantitative analysis of MS lesions in various treatment trials. Most earlier MR studies have been focused on the assessment of MS plaques and their correlations with neurological disability [3,10,14,24]. It was observed that hyperintense lesions on T2-weighted images (T2-plaques) cannot differentiate the separate pathological stages of MS

 1999 Elsevier Science B.V. All rights reserved.

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42

lesion development and represent both active and inactive disease [1]. They display overall lesion load. The degree of hypointensity of MS plaques on T1-weighted images (T1plaques) correlated with the extent of axonal damage, extracellular edema and the degree of demyelination or remyelination [4]. The gadolinium-enhanced plaques on T1-weighted images were found to display the currently active MS plaques [29]. The appearance of new enhancing or non-enhancing lesions or the re-enhancement of the old lesions on MRI was reported to be 5 to 10 times greater than could be expected clinically [30]. The correlations between the number, size, volume and site of MS plaques on conventional MRI examination and neurological disability have been disappointingly weak and only apparent in relapsing-remitting MS [10,15,16]. In general visualization of MS plaques on MRI increased the understanding of the disease, but the objective marker to evaluate disability and the efficacy of the treatment aimed at preventing disability were highly needed. Cerebral and spinal atrophy are important new findings that represent global tissue loss and seem to be correlated with disability measures [11–13,26,27]. Progressive cerebral atrophy correlates with worsening disability when analyzed on conventional MRI technique [5,19,27,34]. This observation indicates that the measurement of atrophy provides an objective marker by which to evaluate the effect of treatment on neurological disability. With the use of new volumetric techniques, estimation of cerebral atrophy is becoming more accurate and less time consuming than the earlier use of linear indices. The purpose of the present study was to correlate the segmented volumes of intracranial cerebrospinal fluid (CSF) spaces representing brain atrophy and the volumes of MS plaques with neurological disability expressed in expanded disability status scale (EDSS) [25] and regional functional scoring system (RFSS) scores [31,35] in secondary progressive MS. Correlations between MRI volumetric measurement and RFSS scores are reported for the first time. All volume determinations were performed by using a new microcomputer applicable semiautomatic MRI segmentation technique.

37

initial symptoms to the beginning of the progressive phase) (range 27–240 months, mean 132612.4 standard error of mean (SEM)) and the progressive phase (the time from the beginning of the progressive phase to the neurological examination, including MRI examination) (range 2–132 months, mean 4966.0 SEM) of the MS disease were correlated to the MRI parameters, EDSS and RFSS. The neurological and MRI examinations were performed within a two-week period. Three experienced neurologists evaluated the patients. They were blinded to the results of MRI. All neurologists underwent training before the actual evaluation. None of the patients received any kind of steroids and the minimum period from the last steroid administration was six months. No changes in clinical condition of patients occurred during the time from administration of steroid to the date of examination.

2.2. Neuroradiological examinations All patients were studied on the same 0.5 Tesla MRI unit (Vectra GE, Wisconsin, USA) to avoid the influence of interscanner variations on the results. The MRI protocol included sagittal T1 (repetition time (TR)5300 ms, echo time (TE)525 ms, number of excitations (NEX)53, field of view (FOV)522 cm, matrix size5160 / 256), axial T1 fast spin echo (FSE) (TR5300 ms, TE520 ms, NEX55, FOV522 cm, matrix size5160 / 256 echo train length (ETL)55), axial T2 (TR52000 ms, TE5150 ms, NEX51, FOV522 cm, matrix size5192 / 224), axial proton density (TR52000 ms, TE5150 ms, NEX51, FOV522 cm, matrix size5192 / 224), axial 2 mm thick three dimensional (3D) T2 FSE (TR52000 ms, TE5150 ms, NEX51, FOV522 cm, matrix size5192 / 224, ETL516), and axial Gadolinium DTPA (administered intravenously as a bolus of 0.2 mmol / kg) enhanced T1 FSE sequences (TR5300 ms, TE520 ms, NEX55, FOV522 cm, matrix size5160 / 256, ETL55). Axial 3D T2 FSE and T1 FSE images were used for segmentation and volumetric analyses. These images were transferred to another workstation equipped with segmentation software.

2.3. Segmentation and volumetric analysis 2. Subjects and methods

2.1. Subjects Twenty-eight patients (14 males and 14 females, age range 35–58 years, mean age 46) with secondary progressive MS underwent a detailed neurological and neuroradiological examination. All studied patients had definitive MS and fulfilled the Poser’s criteria A1 or A2 [33]. Detailed examination of the neurological disability included evaluation of EDSS scores (range 3.5–6.5, mean 4.8) and total RFSS scores (range 5.6–23.7, mean 13.9). The duration of the relapsing–remitting (the time from the

Segmentation and volumetric analysis were performed using the AnatomaticE segmentation software operating in PC /Windows95 E environment [20]. The software applies IARD [21,22] segmentation algorithm. Segmentation procedure is semiautomatic and consists of several imageprocessing methods, such as image enhancement, amplitude segmentation, region growing, manual editing, and decision trees. An intuitive graphical user interface enables the efficient use of image processing and human pattern recognition. The volumetric accuracy of this program was demonstrated by segmenting the T2 weighted MR images of five fluid filled syringes of known volumes (1–20 cm 3 ) and

38

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42

comparing the volumes obtained on MRI with the known volumes of syringes. In this phantom test, five syringes with known volumes of water (1–20 cm 3 ) were imaged using T2 weighted 3D MRI slices. The test demonstrated an relative error of 1.5% [22]. Inter- and intra-observer studies were performed to establish the reproducibility of the volume estimations. The observers were blinded to the clinical findings of the patients. An expert on MS disease and segmentation techniques analyzed the segmentation findings and found that the computer segmentation was accurate enough both for the MS plaque and intracranial CSF space volume estimation, particularly for the peripheral fluid spaces. The measured volumes of lesions and intracerebral fluid spaces were compared and the variability of these results computed. Segmentation and volumetric measurements of hypointense plaques in T1-weighted images and hyperintense plaques on 3D T2-weighted FSE images were performed separately (Fig. 1). Analysis of intracerebral CSF spaces was done from the 3D T2 FSE weighted images. Depending on the proximity of the plaques to the cerebrospinal fluid spaces, the segmentation varied from fully automatic in cases with no subependymal or cortical plaques to varying amount of manual interaction in cases with subependymal or cortical plaques. The time involved in segmenting the intracerebral fluid spaces varied from 1 to 5 min. The CSF volume estimation was however near automatic in most cases. Total intracranial volume was measured by calculating together the volumes of the segmented grey and white matter and intracranial cerebrospinal fluid spaces. Volumetric measurement of total intracranial CSF spaces was done by assessing the total volume of the ventricular and peripheral CSF spaces. Volumetric measurement of relative intracranial CSF spaces was performed by dividing the total intracranial CSF space volume by total brain volume. These intracranial CSF space volumes were considered as markers of cerebral atrophy. The whole procedure of segmentation of both the plaques and intracerebral fluid spaces required from 20 to 30 min depending on the quality of images and quantity of plaques.

2.4. Statistical analysis Spearman’s rank correlations were used to describe the dependence between two variables at a time supported with the P-values. Multiple regression analysis was used to identify explanatory variables amongst MRI measures for neurological disability. The inter- and intra-observer variations were calculated as percentile differences between the measured average volumes and individual measures. The results were expressed as percentage difference between the measured and average volumes. All statistical analyses were made on a microcomputer using the STATISTICA for

Windows program package (Statsoft, Inc., Tusla, OK, USA). The results were expressed as mean6SEM. Due to the multiple comparisons, the criterion for statistical significance was P-value less than 0.01. All P-values between 0.05 and 0.01 were considered as tendency to associate.

3. Results Using AnatomaticE segmentation technique we estimated that the median total lesion volumes of the brains of our MS patients for T2- and T1-weighted plaques were 5.21 cm 3 and 0.60 cm 3 , respectively. In the assessment of brain atrophy we found that the median volume for total intracranial CSF spaces (total volumes of ventricles and peripheral cerebrospinal fluid spaces) was 131.9 cm 3 . The relation of total intracranial CSF space volumes to total brain volume was on average 0.18. In the inter-observer study, sets of MR images of six randomly selected patients were analyzed by three experienced observers (two neurologists and one neuroresearcher) independently and in the intra-observer study, the same neuroradiologist segmented the same set of MR images four times within two weeks interval. For the MS plaque volumes, the results of the inter- and intra-observer studies have been reported in an earlier study [22]. For the cerebral atrophy, the inter-observer study showed a variability of 1.5% between the observers and in the intraobserver study, a variability of 1% was observed between the readings. The correlations between the MRI parameters and neurological disability as expressed by EDSS and RFSS are shown in Table 1. No associations were detected between the MRI volumetric measurements of MS plaques and EDSS. A tendency to association was found between the mean volume of T2 weighted plaques and total RFSS scores (r50.40, P50.03). The associations between the mean volumes of T1 weighted plaques or the volumes of total or relative CSF spaces and total RFSS scores were not significant. No associations were detected between the total brain volumes and EDSS or total RFSS scores. In multiple regression analysis, the volumes of total (P50.006) and relative (P50.005) intracranial CSF spaces were more significantly associated with the neurological disability as expressed by EDSS than other MR measures (Table 2). No associations were found between the total or relative intracranial CSF space volumes and total RFSS scores. The duration of progressive phase of the secondary progressive MS showed a tendency to associate with the EDSS score (r50.37, P50.05). The duration of the progressive phase of MS showed a tendency to associate with the volumes of T1 weighted lesions (r50.38, P5 0.05) but no associations were found with the CSF space volumes or the volumes of T2 weighted lesions. No

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42

39

Fig. 1. Segmentation of multiple sclerosis plaques and atrophy in magnetic resonance imaging (MRI): (A) the high intensity plaques on a T2-weighted image, (B) the gliotic plaques on a T1-weighted image, and (C) the cerebrospinal fluid spaces on a T2-weighted image. Original images on the left and corresponding segmented images on the right. All the segmented structures were obtained using the IARD segmentation algorithm applying appropriate threshold coefficients (T2 plaques; high pass threshold, T1 plaques; low pass threshold, and CSF spaces; high pass threshold).

40

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42

Table 1 Correlations between EDSS, RFSS, and quantitative MRI volumes in 28 patients with secondary progressive multiple sclerosis MRI-parameter volume

EDSS score *R

P-level

RFSS score *R

P-level

Total intracranial CSF space volume, cm 3 Total brain volume, cm 3 Relative intracranial CSF space volume T1-weighted plaque volume, cm 3 T2-weighted plaque volume, cm 3

0.006 20.08 0.03 0.27 0.16

0.97 0.68 0.87 0.15 0.38

0.07 20.15 0.03 0.33 0.40

0.74 0.44 0.86 0.08 0.03

EDSS5expanded disability status scale. RFSS5regional functional scoring system. CSF5cerebrospinal fluid. *Spearmans’s rank order correlation.

correlations were detected between the duration of relapsing–remitting phase of secondary progressive MS and quantitative MRI measurements or neurological disability as expressed by EDSS or RFSS.

4. Discussion In this study, using computerised semiautomatic quantification of MRI parameters in patients with secondary progressive MS we found that an increasing total and relative intracranial CSF volumes indicative of brain atrophy were associated with worsening neurological disability as expressed by EDSS. Progressive cerebral atrophy which represents axonal loss has been reported in patients with MS [27,32]. Axonal injury has been explained by an inflammation [9,39], but recent studies have shown that atrophy may evolve despite the absence of inflammatory activity [28]. It has been shown that during the course of MS corpus callosum (CC) atrophy appears earlier than atrophy in other parts of the brain and its severity correlates with severity of clinical symptoms [8]. Earlier, it was reported that CC atrophy is caused by lesions in and around CC i.e. due to focal areas of demyelination, decrease in number of axons or due to Wallerian degeneration [38]. Later, it was shown that the total area of the CC in MS patients was significantly smaller than in control patients [2]. Progres-

Table 2 Explanatory factors for neurological disability as expressed by EDSS (expanded disability status scale) in multiple regression analysis in patients with secondary progressive multiple sclerosis Explanatory variable

P-value

T1-weighted plaque volumes Total brain volume T2-weighted plaque volumes Relative intracranial cerebrospinal fluid space volume Total intracranial cerebrospinal fluid space volume Entire model

0.067 0.010 0.355 0.005 0.006 0.049

sive cerebral atrophy has been shown to correlate with worsening disability [5,19,34]. However disability may progress even without visible MRI activity of MS lesions [23]. The development of cerebral atrophy early in the course of the disease has been found to be poor prognostic sign for the development of disability [17]. Recent studies have emphasized that the development of brain and spinal cord atrophy occurs especially during progressive phase of the disease and is associated with increasing disability [13,28]. The presence of brain atrophy is more prominent in patients with sustained deterioration of EDSS scores [27]. A positive association has been found between the degree of neurological disability and the extent of periventricular MS lesion load, leading to central brain atrophy [3]. In our study, using MRI and a new fast and simple segmentation technique, the estimation of the cerebral atrophy was more quantitative and accurate than in earlier studies with conventional MRI techniques [8,19] allowing better correlations between atrophy and neurological disability. AnatomaticE is a robust method of quantifying brain atrophy. The software has been found to be comparable to the fastest segmentation techniques available in the market [6]. In addition, the novelty of this software lies in that it can work in any simple PC environment and needs no sophisticated computers. This method may contribute significantly to the evaluation of therapeutic efficacy especially in secondary progressive MS. In this study, the associations between the extent of quantified MS plaques with EDSS remained negative. In several earlier studies, no remarkable associations have been found between the extent of brain abnormalities on conventional MRI examination and neurological disability evaluated by EDSS [14,24,37]. The lack of correlations between the volumetric MS lesion measurements and EDSS in this study is in agreement with previous observations. Positive associations were detected between the volumes of T2 weighted lesion load and total RFSS scores but otherwise use of RFSS did not provide additional information when compared to EDSS. To our knowledge the

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42

correlations between RFSS scores and volumetric measurements of segmented cerebrospinal fluid spaces and MS plaques in secondary progressive MS were performed for the first time in this study. Overall T1-plaque volume measurements correlate better with disability than T2 volume measurements [40]. An explanation for this is given by the recent studies indicating that the degree of hypointensity on T1-weighted images correlates significantly with axonal density and with degree of matrix destruction [41]. In biopsy proven cases, T1-weighted scans revealed major differences in the degree of hypointensity of MS plaques that correlated with the extent of axonal damage, extracellular edema, and the degree of demyelination or remyelination [4]. These observations indicate that the hypointense lesions on T1weighted images may represent more disabling pathology resulting in persistent MS-related neurological deficit. No significant correlations were found between EDSS and hypointense MS plaques in our study. This may be explained by the small sample of patients in this study. Also, measuring hypointense plaques on T1 weighted images is somewhat arbitrary and subjective process which may have hampered the correlations. New quantitative techniques like magnetization transfer imaging should be used for obtaining better results. In this study, the duration of the secondary progressive phase tended to associate with the EDSS. This observation indicates that the severity of the neurological disability increases with the duration of the progressive phase of the MS disease. In conclusion, we believe that our new, simple and fast microcomputer applicable semiautomatic segmentation technique is a sensitive and practical measure of the disability in MS. This study demonstrates that it is possible to quantitate cerebral atrophy and MS plaques in a fast, simple and accurate way using new segmentation techniques. In addition, they can be easily incorporated into various MRI imaging protocols. Our method shall be in future available for commercial use in any neurological workstation for daily clinical practice and during clinical trials of MS.

Acknowledgements The authors thank Pertti Ryymin for skillful technical assistance. The study was supported by Medical Research Fund of the Tampere University Hospital, The Finnish Foundation of Cardiovascular Research, The Elli and Elvi Oksanen Fund of the Pirkanmaa, Regional Fund under the auspices of the Finnish Cultural Foundation, Emil Aaltonen Foundation, Wihuri Foundation, Ulla Tuominen Foundation, Instrumentarium Science Foundation, Ragnar Granit Foundation, The Finnish Radiological Association and Pirkanmaa Medical Fund.

41

References [1] Barkhof F, Frequin STFM, Hommes OR et al. A correlative triad of gadolinium DTPA MRI, EDSS and CSF-MBP in relapsing multiple sclerosis patients treated with high-dose intravenous methylprednisolone. Neurology 1992;42:63–7. [2] Barkhof F, Elton M, Lindeboom J et al. Functional correlate of calloasl atrophy in relapsing2remitting multiple sclerosis patients. A preliminary MRI study. J Neurol 1998;245:153–8. [3] Baum K, Nehrig C, Girke W, Brau H, Schorner W. Multiple sclerosis: relations between MRI and CT findings, cerebrospinal fluid parameters and clinical features. Clin Neurol Neurosur 1990;92:49–56. [4] Bruck W, Bitsch A, Kolenda H et al. Inflammatory central nervous system demyelination: correlation of magnetic resonance imaging findings with lesion pathology. Ann Neurol 1997;42:783–93. [5] Comi G, Filippi M, Martinelli V et al. Brain MRI imaging correlates of cognitive impairment in multiple sclerosis. J Neurol Sci 1993;115(suppl):S66–73. [6] Dastidar P, Heinonen T, Vahvelainen T, Elovaara I, Eskola H. Computerised volumetric estimation of lesions in multiple sclerosis using a new semiautomatic segmentation software. Med Biol Eng Comput 1999;37:104–7. [7] DeCarli C, Maisog J, Murphy DGM et al. Method for quantification of brain, ventricular, and subaracnoid CSF volumes from MR images. J Comput Assist Tomogr 1992;16:274–84. [8] Dietemann JL, Beigelman C, Rumbach L et al. Multiple sclerosis and corpus callosum atrophy: relationship of MRI findings to clinical data. Neuroradiology 1988;30:478–80. [9] Fergusson B, Matyszak MK, Esiri MM, Perry VH. Axonal damage in acute multiple sclerosis lesions. Brain 1997;120:393–9. [10] Filippi M, Paty DW, Kappos L et al. Correlations between changes in disability and T2 weighted brain MRI activity in multiple sclerosis. A follow up study. Neurology 1995;45:478–80. [11] Filippi M, Mastronardo G, Rocca MA, Pereira C, Comi G. Quantitative volumetric analysis of brain magnetic resonance imaging from patients with multiple sclerosis. J Neurol Sci 1998;158:148–53. [12] Filippi M, Colombo B, Rovaris M, Pereira C, Martinelli V, Comi G. A longitudinal magnetic resonance imaging study of the cervical cord in multiple sclerosis. J Neuroimaging 1997;7:78–80. [13] Filippi M, Campi A, Colombo B et al. A spinal cord MRI study of benign and secondary multiple sclerosis. J Neurol 1996;243:502–5. [14] Franklin GM, Heaton RK, Nelson LM, Filley CM, Seibert C. Correlation of neuropsychological and MRI findings in chronic / progressive multiple sclerosis. Neurology 1988;38:1826–9. [15] Gasperini C, Horsfield MA, Thorpe JW et al. Macroscopic and microscopic assessment of disease burden in multiple sclerosis: relationship to clinical parameters. J Magn Reson Imag 1996;6:580– 4. [16] Gass A, Barker GJ, Kidd D et al. Correlation of magnetic transfer ratio with clinical disability in multiple sclerosis. Ann Neurol 1994;36:62–7. [17] Gross KR, Tomberg TA, Kokk AA et al. The prognosis of multiple sclerosis: computed tomography comparisons. Zh Nevropatol Psikhiatr Im S S Korsakova 1993;93:32–5. [18] Grossman RI, Gonzales-Scarano F, Atlas SW, Galetta S, Siberberg DH. Multiple sclerosis: gadolinium enhancement in MR imaging. Radiology 1986;161:721–5. [19] Hageleit U, Will CH, Seidel D. Automated measurements of cerebral atrophy in multiple sclerosis. Neurosurg Rev 1987;10:137–40. [20] Heinonen T, Eskola H, Dastidar P et al. Segmentation of T1 MR scans for reconstruction of resistive head models. Comput Meth Prog Bio 1997;54:173–81. [21] Heinonen T, Dastidar P, Malmivuo J, Eskola H. Semiautomatic tool for the segmentation and volumetric analysis of medical images. Med Biol Eng Comput 1998;36:291–6. [22] Heinonen T, Dastidar P, Eskola H et al. Applicability of

42

[23]

[24]

[25]

[26]

[27] [28]

[29] [30]

[31] [32] [33]

P. Dastidar et al. / Journal of the Neurological Sciences 165 (1999) 36 – 42 semiautomatic segmentation for volumetric analysis of brain lesions. J Med Eng Technol 1998;22:173–8. Kidd D, Thorpe J, Barker GJ et al. MRI dynamics of the brain and spinal cord in progressive multiple sclesrosis. J Neurol Neurosurg Psychiatry 1996;60:15–9. Kiel MK, Greenspun B, Grossman RI. Magnetic resonance imaging and degree of disability in multiple sclerosis. Arch Phys Med Rehab 1988;69:11–3. Kurtzke JF. Rating neurologic impairment in multiple sclerosis. An expanded disability status scale (EDSS). Neurology 1983;33:1444– 52. Loseff NA, Webb SL, O’Riordan JI et al. Spinal cord atrophy and disability in multiple sclerosis: a new reproducible and sensitive MRI method with potential to monitor disease progression. Brain 1996;119:701–8. Loseff NA, Wang L, Lai HM et al. Progressive cerebral atrophy in multiple sclerosis. A serial MRI study. Brain 1996;119:2009–19. Loseff NA, Miller DH. Measures of brain and spinal cord atrophy in multiple sclerosis. J Neurol Neurosurg Psychiatry 1998;64(Suppl. 1):102–5. McDonald WI, Barnes D. Lessons from magnetic resonance imaging in multiple sclerosis. Trends Neurosci 1989;12:376–9. Miller DH, Rudge P, Johnson G et al. Serial gadolinium enhanced magnetic resonance imaging in multiple sclerosis. Brain 1989;111:927–39. Munsat T, editor, Quantification of neurological deficit, Butterworth, Boston, 1989, pp. 53–4. Ozawa K, Suchanek G, Breitschopf H et al. Patterns of oligodendroglia pathology in multiple sclerosis. Brain 1994;117:1311–22. Poser CM, Paty DW, Scheinberg L et al. New diagnostic criteria for

[34]

[35]

[36]

[37]

[38]

[39] [40]

[41]

[42]

multiple sclerosis: Guidelines for research protocols. Ann Neurol 1983;13:227–31. Rao SM, Glatt S, Hammeke TA et al. Chronic progressive multiple sclerosis. Relationship between cerebral ventricular size and neuropsychological impairment. Arch Neurol 1985;42:678–82. Runmarker B, Andersen O. Prognostic factors in a multiple sclerosis incidence cohort with twenty years of follow-up. Brain 1993;116:117–34. Rydberg JN, Hammond CA, Grimm RC et al. Initial experience in MR imaging of the brain with fast fluid attenuated inversion recovery pulse sequence. Radiology 1994;193:173–80. Simons JH, Jacobs LD, Campion M, Wende K, Simonian N. Magnetic resonance studies of intramuscular Interferon B-1a for relapsing multiple sclerosis. Ann Neurol 1998;43:79–87. Simons JH, Holtas SL, Schiffer RB et al. Corpus callosum and subcallosal periventricular lesions in multiple sclerosis: detection with MR. Radiology 1986;160:363–7. Trapp BD, Peterson J, Ranshoff RM et al. Axonal transection in lesions of multiple sclerosis. N Eng J Med 1998;338:278–85. Truyen L, VanWaesberge JHTM, Barkhof F et al. Accumulation of hypo-intense lesions (black holes) on T1 weighted SE MRI in multiple sclerosis correlates with disease progression. Neurology 1996;47:1469–76. vanWalderveen MAA, Barkhof F, Hommes OR et al. Correlating MR imaging and clinical disease activity in multiple sclerosis: relevance of hypointense lesions on short TR / short TE (T1 weighted) spin echo images. Neurology 1995;45:1684–90. Wolf SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue proton relaxation in vivo. Magn Reson Med 1989;10:135–44.