METHODS

The design, conduct and clinical results of the trial have been reported previously.⁷ Briefly, 66 patients with a first episode of acute unilateral optic neuritis within 30 days of onset were enrolled into the study. The median duration of symptoms before randomisation was eight days (range 1–30). Six of the patients had clinically definite MS, another 14 had clinically probable MS, and the rest had clinically isolated optic neuritis. Their optic nerves were imaged with a short tau inversion recovery (STIR) sequence (TR 2500 ms, TE 40 ms, TI 175 ms, matrix 256×128, field of view 16 cm × 16 cm, 2 excitations, 5 mm contiguous slices) and were then randomised to receive either 1 g/day IVMP for three days or intravenous saline. Reimaging was performed six months later. In addition, at six months, a detailed clinical assessment was performed including Snellen visual acuity, contrast sensitivity using the Pelli-Robson chart, 30-2 Humphrey visual field examination, and colour vision using the Farnsworth-Munsell 100 (FM 100) Hue test. Normal values were taken to be a visual acuity of 6/6 or better, contrast sensitivity of 1.65 or better, Humphrey mean deviation of −3 dB or higher, and the total error score of the FM 100 of less than 110.^8,9 Whole field and central field pattern evoked visual evoked potentials (VEP) were also recorded.

Only images that were acquired on a Signa 1.5 T imager (General Electric, Milwaukee, WI) were available. Some early patients in the study were imaged on a Picker 0.5 T imager and these images were not available for study. In total, images from 45 patients at baseline and 59 patients after six months (30 given IVMP and 29 given placebo) were examined. The images were transferred onto workstations (Sun Microsystems, Mountain View, CA). Mean optic nerve area was measured by an observer blinded to image identity, treatment group, and acquisition order from two consecutive 5 mm orbital slices from the orbital apex forwards using a semiautomated contouring technique as described previously.⁶ Each measurement was carried out three times independently of each other. In a previously reported study of a different patient cohort, five consecutive 3 mm orbital slices from a short echo fast fluid attenuated inversion recovery (sTE fFLAIR) sequence were measured.⁶ It was not possible to measure a 15 mm segment in all subjects from the STIR images as in some patients the most anterior slice occurred at the point that the nerve sheath dilated as it attached on the back of the globe leading to an artificially increased area; hence a 10 mm segment was measured. This was less of a problem with the previous study because the sTE fFLAIR sequence suppresses the signal from cerebrospinal fluid. The presence of high signal lesions in the measurement area was noted and lesion lengths at baseline and six months were recorded from the data of the original study, measured by an experienced neuroradiologist.⁷

Statistical methods

The presence of nerve swelling at baseline and atrophy at six months was assessed in terms of the ratio diseased nerve area:healthy nerve area at the respective time points. Null hypotheses of ratio = 1 were examined by paired t tests on log nerve areas. The possible influence of a lesion in the measured portion of the nerve was examined by regression of log ratios on the presence of a lesion and its length. The association between atrophy and steroid use was assessed by regression of the log ratio of six month diseased nerve area:six month healthy nerve area and six month diseased area:baseline diseased nerve area on an indicator of steroid use, adjusting for relevant baseline nerve area. The influence of the presence of a baseline lesion was assessed by an interaction term. All paired tests or analyses involving both time points used only patients available at both time points. Associations between diseased nerve area at six months and VEP variables were assessed by linear regression on log diseased nerve area at six months; clinical variables were dichotomised about the threshold for the normal ranges and similarly assessed by logistic regression. To assess measurement reproducibility the within and between subject standard deviations (and hence coefficient of variation and reliability coefficients) were obtained from random effects analysis of variance.¹⁰

RESULTS

Table 1 gives measurement reproducibility results. At baseline optic nerve mean area was 18.4 (SD 3.8) mm² in diseased optic nerves and 17.8 (SD 3.6) mm² in healthy optic nerves (n = 45). The estimated geometric mean ratio (diseased nerve area:healthy nerve area) was 1.035 (95% CI 0.96 to 1.11; p = 0.33). At baseline, high signal was present in the measurement area in 36/45 patients. The presence of a lesion did not affect the ratio; however, the degree of swelling increased by 7.5% (95% CI 3.3% to 11.7%; p = 0.001) for each slice that a high signal lesion was visible on.

After six months optic nerve mean area was 16.4 (SD 3.8) mm² in diseased optic nerves and 17.4 (SD 3.5) mm² in healthy optic nerves (n = 59) (fig 1). The estimated geometric mean ratio (diseased nerve area:healthy nerve area) was 0.93 (95% CI 0.87 to 0.99; p = 0.02). A lesion was present in 56/59 patients in the orbital portion measured at this time point. Neither the lesion length at baseline or six months correlated with the degree of atrophy. There was no evidence of association between any of the clinical and VEP variables and six month diseased nerve area.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Short tau inversion recovery magnetic resonance images showing optic nerve swelling at baseline (left) and optic nerve atrophy after six months (diseased nerve arrowed) in a 27 year old man with right optic neuritis.

The mean area of affected optic nerves at six months in the IVMP group was 15.9 (SD 3.9) mm² (n = 30) compared with 16.9 (SD 3.8) mm² in the placebo group (n = 29). The ratio of six month diseased:six month healthy optic nerve area was 6.8% lower in the IVMP group than in the placebo group (95% CI, 16.1% lower, 3.6% higher, p = 0.19). The ratio of six month diseased nerve area:baseline diseased nerve area was 0.8% lower in the IVMP group than in the placebo group (95% CI 14.8% lower, 15.5% higher, p = 0.92). Neither the MS status nor the duration of symptoms before treatment was instigated affected these ratios.

DISCUSSION

This technique was able to show optic nerve atrophy following optic neuritis with good reproducibility as witnessed by the high reliability coefficients. The area measurements are greater than those produced from sTE fFLAIR images as the measurements from the STIR images include the nerve sheath. Even though the measurements were of the optic nerve–sheath complex, atrophy was still detected after six months. At the time that the images were acquired during the trial, the sTE fFLAIR sequence had not been developed and STIR was the preferred sequence for optic nerve lesion identification.

A study using sTE fFLAIR in a more chronic cohort of optic neuritis patients showed that increasing optic nerve atrophy was associated with worse vision and decreased VEP amplitudes.¹¹ Qualitative assessment of the sTE fFLAIR images from that study suggests that atrophy of the optic nerve sheath occurred as well (SJ Hickman, unpublished observations). The lack of correlation between the clinical outcome measures and optic nerve mean area in the present study may be caused by functional reorganisation in the visual system in the early recovery process,¹² potentially achieved by utilising redundant optic nerve capacity.¹³ This plasticity may fail over time and this may be one explanation for the development of late clinical progression in MS.

There is no evidence from these data that a course of IVMP prevents the short term development of optic nerve atrophy following acute optic neuritis. This is consistent with the lack of long term functional benefit seen as a result of IVMP in both this and other studies.