Upper cervical cord area in early relapsing-remitting multiple sclerosis: Cross-sectional study of factors influencing cord size

Authors

  • Waqar Rashid PhD,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Gerard R. Davies PhD,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Declan T. Chard MB, BS,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Colette M. Griffin MD,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Dan R. Altmann DPhil,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    2. Medical Statistics Unit, London School of Hygiene and Tropical Medicine, London, United Kingdom
    Search for more papers by this author
  • Ros Gordon MSc,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Raju Kapoor MD,

    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • Alan J. Thompson MD,

    1. MS NMR Research Unit, Department of Headache, Brain Injury and Rehabilitation, Institute of Neurology, University College London, London, United Kingdom
    Search for more papers by this author
  • David H. Miller MD

    Corresponding author
    1. MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, London, United Kingdom
    • MS NMR Research Unit, Department of Neuroinflammation, Institute of Neurology, University College London, Queen Square, London WC1N 3BG, UK
    Search for more papers by this author

Abstract

Purpose

To determine whether the upper cervical cord area (UCCA) is influenced by disease effect in early relapsing-remitting multiple sclerosis (MS), using statistical modeling to account for potential covariates.

Materials and Methods

A cohort of 39 patients were studied cross-sectionally within three years of first symptom onset (median disease duration = 1.6 years) and compared with 26 healthy controls. The UCCA was measured from axial reconstructions of three-dimensional T1-weighted scans with automated detection of the edge of the cord. Statistical analysis adjusted for factors such as total intracranial volume (TICV) and gender. Clinical correlations, in particular those thought likely to be related to cord pathology, were also investigated.

Results

No significant disease effect was noted on UCCA (P = 0.685), although there was borderline evidence of a lower UCCA in patients with symptoms of bowel or bladder disturbance (P = 0.043). A strong association was noted between UCCA and TICV (r = 0.558; P ≤ 0.001), and there was a trend for females to have a smaller UCCA (P = 0.062). The latter finding appeared to reflect a gender-related difference in TICV (P ≤ 0.001).

Conclusion

Atrophy of the upper cervical cord is not readily apparent in most patients early in the course of relapsing-remitting MS. In evaluations of disease-related changes in the UCCA in cross-sectional studies, TICV and gender should be considered as potentially confounding covariates. J. Magn. Reson. Imaging 2006. © 2006 Wiley-Liss, Inc.

INVOLVEMENT OF THE SPINAL CORD in multiple sclerosis (MS) is common and relevant in the development of disability (1). Magnetic resonance imaging (MRI) offers a means of measuring atrophy resulting from this disorder in vivo, and a number of studies have revealed significant cord atrophy cross-sectionally in patients with progressive forms of MS (2–13). Further, a strong correlation between cervical cord atrophy and disability was found in some studies (2, 3, 6, 8).

It is unclear how early in the course of MS cervical cord area loss starts to occur. A cross-sectional study in patients with clinically isolated syndromes (CIS) suggestive of MS demonstrated cord area loss in subjects with abnormal MRI findings, but showed no longitudinal changes over one year (14). However, most other cross-sectional studies have showed no upper cervical cord area (UCCA) differences between relapsing-remitting MS and controls (2, 5, 6, 9, 10), although two studies did observe significant longitudinal loss in patients after one year (5) and four years (6), respectively.

The somewhat variable findings previously reported in CIS and relapsing-remitting MS may reflect differences in cohort recruitment, cohort size, and disease duration. Variations in the total intracranial volume (TICV) and gender may also be relevant since the TICV has been correlated with the UCCA (9, 10, 15), and also differs between males and females (16). The aim of this study was to determine whether UCCA loss could be detected cross-sectionally in a sizeable cohort of clinically definite relapsing-remitting MS subjects at an early stage of the disease, using a sensitive and established technique (2) and accounting for TICV and gender in the analysis.

MATERIALS AND METHODS

Thirty-nine patients (28 females and 11 males, mean age = 36 years, range = 24.1–55.6 years) with clinically definite relapsing-remitting MS were recruited within three years of symptom onset (median disease duration = 1.6 years, range = 0.5–3 years, median expanded disability status scale (EDSS) (17) = 1.0, range = 0–3) and compared with 26 healthy volunteers (15 females and 11 males, mean age = 35.8 years, range = 23.2–52.7 years). None of the subjects had ever commenced beta-interferon treatment, or experienced a clinical relapse or received a course of corticosteroids within the previous month. The EDSS and MS Functional Composite Measure (MSFC) (18) were assessed in each patient. Informed consent was obtained from all subjects and the study was approved by the joint medical ethics committee of our institution.

Scans were performed on a 1.5 Tesla GE Signa Horizon Echospeed scanner (General Electric Medical Systems, Milwaukee, WI, USA) running on 5.x software. All subjects underwent imaging of the spinal cord with a volume-acquired inversion-prepared fast spoiled gradient echo (bandwidth = 10.42; TR = 15.6 msec; TE = 4.2 msec; TI = 450 msec; matrix = 256 × 256; FOV = 250 × 250; partial echo) forming 60 × 1 mm sagittal slices covering the whole of the cervical cord. From this, five contiguous 3-mm pseudoaxial slices were reformatted using the center of the C2/C3 intravertebral disc as a caudal landmark, with the slices perpendicular to the spinal cord. In the same scanning session, after the volume sequence was acquired, triple-dose gadolinium (Gd)-DTPA (0.3 mmol/kg) was injected, and approximately 15 minutes later a conventional spin-echo T1-weighted sequence of the cord was acquired (bandwidth = 20.83; TR = 500 msec; TE = 20 msec; matrix = 512 × 512; FOV = 480 × 240; full echo) forming 13 × 5 mm sagittal slices of the cord. A 3D inversion-prepared fast spoiled gradient recall sequence (bandwidth = 15.63; TR = 13.3 msec; TE = 4.2 msec; TI = 450 msec; matrix = 256 × 144; FOV = 240 × 180 mm [zero-filled to give an image pixel size of 1.2 × 1.2 mm]; partial echo) providing 124 × 1.5 mm slices covering the whole cranium was also obtained during a separate scanning session (median = 7 days after cord imaging).

These images were transferred to a Sun workstation (Sun Microsystems Inc., Mountain View, CA, USA) and uniformity corrected. With the use of an automated program, the mean area of the cord was calculated from the five pseudoaxial cervical cord images with a technique described by Losseff et al (2), with the observer blinded to clinical status. Gd-enhancing lesions of the cord were identified from the postcontrast images of the cord by an experienced observer and outlined on the images using the Dispimage display software package (Plummer, Department of Physics, University College Hospitals NHS Trust, London, UK) and the volumes were calculated. The TICV was calculated from the 3D cranial volume images using statistical parametric mapping (SPM99, Wellcome Department of Cognitive Neurology, Institute of Neurology, London, UK) as previously described (19).

In nine subjects (five patients and four controls) the UCCA measurements were repeated twice on the same images 30 days apart in a blinded fashion to assess the reproducibility of the observer's measures. The coefficient of variation was calculated from these results.

Statistical analysis was performed using SPSS 11.0 (SPSS Inc., Chicago, IL, USA). The univariate association between TICV and UCCA was investigated using Pearson correlation analysis. Unpaired t-tests were used for univariate tests of UCCA difference between genders, and between patients and controls. Multiple linear regression models were used to determine the potential effect of age, gender, TICV, disease duration (in patients only), and subject weight on UCCA. This allowed us to calculate the potential disease effect (in MS patients and controls) on the UCCA while controlling for the above variables. Linear regression with these covariates was also used to determine any association between MSFC and UCCA. Two-tailed Spearman rank correlation coefficients were used to assess any relationship between the UCCA and the EDSS. Regression of the UCCA on the functional systems contained within the EDSS, which may be more specific to spinal cord pathology (e.g., pyramidal, brainstem, bowel, or bladder symptoms, or sensory disturbance), was performed for affected and unaffected patients, with TICV used as a covariate. A P-value < 0.05 was considered significant.

RESULTS

Table 1 shows the UCCA and TICV for controls and patients. When TICV and gender were adjusted for, there was no significant UCCA loss in the MS subject group compared to controls (P = 0.685; mean difference controls – patients = 0.76 mm2 [95% confidence interval (CI) = –2.98 to 4.5]). Similarly, when these covariates were not adjusted for, no difference between patients and controls was observed (P = 0.671; mean difference controls – patients = 0.94 mm2 [95% CI = –3.94 to 5.31]). The UCCA in both patient and control cohorts had a similar spread (SD patients = 9.02; SD controls = 8.04), and no patients were identified with a UCCA more than 2 SD below the mean normal control UCCA.

Table 1. Upper Cervical Cord Area and Total Intracranial Volume of Each Cohort*
 MS subjects (N = 39)Controls (N = 26)
  • *

    Mean values (SD) unless stated.

Cord area (mm2)78.62 (9.02)79.56 (8.04)
 Males80.60 (5.63)82.96 (8.94)
 Females77.85 (10.03)77.07 (6.53)
Total intracranial volume (mL)1339.70 (160.26)1334.97 (141.90)
 Males1475.19 (156.44)1446.90 (137.23)
 Females1291.31 (133.45)1255.03 (77.91)

No significant association was noted in patients between the UCCA and MSFC, and no correlation was seen between the EDSS and UCCA (P = 0.171). Regarding functional systems suggestive of cord pathology, in patients (N = 6; mean UCCA = 72.08 mm2; SD = 5.74) with bowel or bladder symptoms (defined as a score of at least one in this functional system), a borderline significantly lower UCCA was noted in comparison to controls (P = 0.055) and other patients (N = 33; mean = 79.81 mm2; SD = 9.06) without such symptoms (P = 0.043). However, after the TICV was adjusted for, the significance decreased (P = 0.133 and P = 0.089, respectively).

In all subjects combined, univariate tests showed a strong positive association between TICV and UCCA (r = 0.558; P ≤ 0.001; see Fig. 1), while there was a borderline significant decrease of UCCA in females (P = 0.062; mean difference male – female = 4.2 mm2 [95% CI = –0.21 to 8.6]). Similar correlations were observed for controls only (TICV and UCCA [r = 0.461; P = 0.023] and gender and UCCA [P = 0.064]), nor was there any evidence of a difference in these associations between patients and controls. There was strong evidence that females have a lower TICV than males (P ≤ 0.001; mean difference male – female = 181.83 mL [95% CI = 112.88–250.77]). When the TICV gender difference was adjusted for, the trend for gender to effect cord area was lost (P = 0.260; mean difference male – female = –2.61 mm2 [95% CI = –7.2 to 1.97]). No other significant covariates were found.

Figure 1.

Scatter graph showing the correlation between the UCCA and TICV for all subjects, with fitted regression line (Pearson r = 0.57; P ≤ 0.001).

Four subjects had cervical cord Gd-enhancing lesions in the area corresponding to the section used to calculate the UCCA. No significant difference was seen between these four subjects (mean UCCA = 78.96, SD = 8.79) and the remainder of the patient or control groups (mean UCCA = 78.59, SD = 9.17). In addition, removing the four subjects with Gd-enhancing lesions in the UCCA region of interest (N = 35) caused no significant difference in the overall comparative analysis with controls. The single-operator, same-image scan reproducibility of the nine subjects' data was 1.06%.

DISCUSSION

The results of this study suggest that the main influences on UCCA in early relapsing-remitting MS are the TICV and gender, rather than any potential disease effect. In addition, the effect of gender appears to be mediated by the TICV, with no independent association with the spinal cord itself (as illustrated by the reduced and nonsignificant adjusted gender effect). The apparent absence of a disease effect on the UCCA is reinforced by the finding of no difference between patients and controls when they were analyzed without adjustment for covariates, even though there was a higher ratio of males in the control group.

The present analysis revealed no significant atrophy in patients with early relapsing-remitting MS. Given the correlation shown in many studies between the UCCA and cord volume to the EDSS (2–11, 13), and the possible association between disease duration and cord atrophy discussed in other studies (8, 13), the lack of significant atrophy in the present cohort of relapsing-remitting MS is consistent with the patients' short disease duration and mild clinical impairment.

A number of previous studies that investigated the UCCA or cord volumes in MS suggested an association between the calculated cord values and TICV (6–10), and attempted to correct for this influence in their analyses. However, they did not explore the possible relationship between gender and TICV. Given the strong association between gender status and TICV, and the consequent relationship shown in this study among TICV, gender status, and the UCCA, the present results illustrate the importance of adjusting for both TICV and gender (particularly if TICV is unavailable) in any UCCA or cord volume analysis in MS. In previous studies, subject height was suggested to have a possible relationship with the UCCA (2, 15); however, the cohort information from those studies was not available for the present analysis. Hence, we are unable to exclude the possibility that height differences between male and female subjects may partially mediate gender UCCA differences. In contrast to a previous study (2), we did not find subject weight to have a significant influence on the UCCA. In keeping with all other studies that have assessed cord volume or area (2–13), subject age was not found to be a significant covariate. In addition, the presence of Gd-enhancing lesions in the cord region studied was not found to influence the area; however, the number of patients who exhibited such lesions was small (N = 4). It should also be noted that another possible covariate, non-enhancing T2-weighted lesions in the cord, could not be investigated in this study because T2-weighted images of the cord were not obtained.

Previous studies that investigated CIS and relapsing-remitting MS cord size produced different results. Most cross-sectional studies showed no significant cord atrophy in relapsing-remitting MS cohorts (all with greater median disease durations than the cohort investigated in this work) compared to controls (2, 5–7, 9, 10); however, some evidence of progressive atrophy was produced by longitudinal studies (5, 6). A magnetization transfer ratio imaging study in relapsing-remitting MS subjects compared to normal controls also showed no significant differences in the cord (20). Conversely, a study that investigated the UCCA in CIS showed significant baseline atrophy in patients with abnormal MRI scans compared to controls (14). Further, a study that measured cord volume showed a significant decrease in relapsing-remitting MS patients compared to controls (12). However, neither of these investigations performed gender correction between the cohorts, and the CIS analysis (14) did not adjust for TICV. The present study provides further evidence, using a robust methodology, that significant UCCA atrophy is not apparent during the earliest clinical stages of relapsing-remitting MS.

Progressive MS patients have consistently shown cord atrophy in comparison with normal controls (2–13). The correlation between atrophy and clinical disability (EDSS) reported in some investigations (2–11, 13), and the association between diffuse cord abnormalities and atrophy (4) suggest that findings of cord atrophy in clinical phenotypes of MS could be used as a surrogate marker of disease severity and prognosis, as well as in disease-modifying drug treatment trials. Further, the finding of cord atrophy in patients with secondary progressive MS in comparison to benign subjects (3) suggests that even in patients with long disease duration, cord atrophy has the potential to be a marker of disease severity.

In the present analysis, subjects with urinary or bowel symptoms showed some evidence of cord atrophy; however, these results must be interpreted with caution because the difference in the UCCA was small and of borderline statistical significance. A follow-up of this cohort is under way to determine the long-term evolution of UCCA. It will be relevant to know when disease-related cord atrophy emerges, and whether it is related to clinical evolution and symptoms suggestive of cord pathology, or has a prognostic role in the development of progressive disease or irreversible disability.

In conclusion, we found no significant UCCA in the present cohort with clinically early relapsing-remitting MS. This study does, however, emphasize that in evaluations of disease-related changes in the UCCA, TICV and gender (particularly the former) should be considered as potentially confounding covariates. This should be taken into account in cross-sectional studies investigating the UCCA in any pathological process known to affect the cervical cord.

Ancillary