Twenty-five subjects, all of whom had at least 6 months of follow-up, contributed 31 acutely injured optic nerves (). Three subjects presented with simultaneous and bilateral optic neuritis, and another 3 subjects had sequential optic neuritis, with the onset of optic neuritis occurring in the second eye within 21 days of it occurring in the first eye. In all sequential cases, the second eye had been clinically and radiographically documented as normal at the time the first eye was affected. Fifteen subjects presented with clinically isolated syndrome (optic neuritis with an abnormal brain MRI consistent with demyelination), 5 had isolated optic neuritis (normal brain MRI and negative for neuromyelitis optica IgG antibodies), 1 developed chronic recurrent inflammatory optic neuritis (normal brain and spine MRI and negative for neuromyelitis optica IgG antibodies), and 1 had acute disseminated encephalomyelitis (also with encephalopathy and transverse myelitis, abnormal brain MRI at onset, negative for neuromyelitis optica IgG antibodies, normal brain MRI, and no relapse at 3.5-year follow-up). Three eyes of patients who presented with acute vision loss and an abnormal brain MRI were months later determined to have had nonarteritic ischemic optic neuropathy. Eyes with incomplete recovery from nonarteritic ischemic optic neuropathy did not differ in mean DTI parameters at onset from those with acute optic neuritis (fractional anisotropy [P = .77], mean diffusivity [P= .84], AD [P= .97], and radial diffusivity [P = .81]); analyses were performed with and without inclusion of subjects with nonarteritic ischemic optic neuropathy.
Baseline Characteristics of 25 Subjects Who Presented Within 31 Days After Acute Visual Symptoms Consistent With Optic Neuritisa
LOW BASELINE AD AS A PREDICTOR OF INCOMPLETE VISUAL RECOVERY AT 6 MONTHS
Baseline AD differentiated subjects with complete clinical recovery by 6 months (1.75 μm2/ms [95% CI, 1.67–1.83 μm2/ms]) from those with incomplete recovery by 6 months (1.50 μm2/ms [95% CI, 1.36–1.64 μm2/ms]). Odds of complete 6-month recovery decreased by 53% (95% CI, 27%–70%) for every 0.1-unit decrease in AD at baseline. When both baseline AD and visual acuity were included as covariates in a predictive model of 6-month recovery, AD continued to provide an independent prognostic value (P=.03). Additional covariates in the model for predicting outcome were not significant, including age (P=.58), sex (P=.68), and use of intravenous glucocorticoids (P=.39). After excluding the 3 subjects who were later diagnosed with nonarteritic ischemic optic neuropathy, we found that a low AD continued to predict 6-month incomplete recovery (odds ratio, 0.47 [95% CI, 0.18–0.65]; P = .03). Thus, AD was a predictor of outcome in subjects with inflammatory and demyelinating optic neuritis.
Affected eyes with incomplete recovery demonstrated a lower mean AD (1.50 μm2
/ms [95% CI, 1.36–1.64 μm2
/ms]) than did unaffected fellow eyes (1.79 μm2
/ms [95% CI, 1.67–1.91 μm2
/ms]). Affected eyes with complete recovery (1.75 μm2
/ms [95% CI, 1.67–1.83 μm2
/ms]) were not different from the unaffected fellow eyes. Both incomplete and complete recovery AD values were within 2 SDs of the mean (SD) AD value for healthy controls, 1.66 (0.18) μm2
/ms (interscan SD, 0.12; intrascan SD, 0.13).16
CORRELATION BETWEEN LOW BASELINE AD AND WORSE CLINICAL OUTCOME MEASURES AT 6 MONTHS
Baseline AD correlated with the 6-month visual acuity (r=0.40, P=.03), 5% contrast sensitivity (r=−0.45, P=.01), and Pelli-Robson contrast sensitivity (r = 0.41, P = .02) (). When evaluating only those cases of optic neuritis with moderate to severe onset (visual acuity, ≤0.5), we found that AD continued to correlate with the 6-month 5% contrast sensitivity (r=0.43, P=.05) and Pelli-Robson contrast sensitivity (r = 0.48, P = .03). In addition, lower values of baseline AD correlated with lower VEP amplitudes (r = 0.55, P <.01), thinner RNFLs (r=0.57, P=.02), and the prolongation of VEP latencies (r=−0.38, P=.04) at 6 months. Mean diffusivity demonstrated more variable correlations with 6-month clinical outcomes ().
Correlations Between Baseline and 6-Month Outcome Measuresa
POOR VISUAL ACUITY AT ONSET AS A PREDICTOR OF WORSE RECOVERY
Visual acuity at onset was not predictive of visual acuity at 6 months (r=0.31, P=.94). In the aggregate, baseline visual acuity was a predictor of 6-month Pelli-Robson contrast sensitivity (r=0.43, P=.02), although for the 21 subjects with moderate to severe loss of vision at presentation (visual acuity, ≤0.5), visual acuity was not a predictor of Pelli-Robson contrast sensitivity (r=0.03, P=.91). For the entire group, visual acuity at baseline was not a predictor of 5% contrast sensitivity at 6 months (r=−0.16, P=.39).
CORRELATION BETWEEN LOW AD AND WORSE BASELINE CLINICAL MEASURES
Axial diffusivity at baseline had moderate correlations with baseline visual acuity (r=0.49, P<.01), 5% contrast sensitivity (r=−0.40, P=.02), and Pelli-Robson contrast sensitivity (r=0.50, P<.01) (). Mean diffusivity had similar baseline correlations ().
TIME COURSE FOR LOW AD
Low AD was confined to the first month. Axial diffusivity for the group was 1.58 μm2/ms (95% CI, 1.50–1.66 μm2/ms) over the first month, increasing at 3 months to 1.85 μm2/ms (95% CI, 1.74–1.92 μm2/ms), trending upward at 6 months to 1.93 μm2/ms (95% CI, 1.84–2.03 μm2/ms), and plateauing at 12 months to 1.98 μm2/ms (95% CI, 1.87–2.08 μm2/ms) ().
Figure 1 Time course of changes in the parameters used in diffusion tensor imaging (DTI) of the optic nerves of subjects from onset of optic neuritis during a 1-year period. All cases are included (n = 25). All parameters become altered by 3 months in comparison (more ...)
After dividing subjects by recovery status, we found that the incomplete recovery group had a lower mean baseline AD (1.50 μm2/ms [95% CI, 1.36–1.64 μm2/ms]) compared with the complete recovery group (1.75 μm2/ms [95% CI, 1.67–1.83 μm2/ms]) (). Despite starting from a lower AD value, those with severe injury displayed a greater increase in AD over the ensuing year (0.57 μm2/ms per year [95% CI, 0.42–0.73 μm2/ms per year]) compared with those with full recovery (0.12 μm2/ms per year [95% CI, 0.03–0.21 μm2/ms per year]). Differences in mean AD between recovery groups were not observed at 3 months (1.84 μm2/ms [95% CI, 1.71–1.98 μm2/ms] for the incomplete group vs 1.80 μm2/ms [95% CI, 1.71–1.88 μm2/ms] for the complete group), 6 months (1.94 μm2/ms [95% CI, 1.80–2.08 μm2/ms]) for the incomplete group vs 1.90 μm2/ms [95% CI, 1.80–1.99 μm2/ms] for the complete group), or 12 months (2.03 μm2/ms [95% CI, 1.89–2.17 μm2/ms] for the incomplete group vs 1.89 μm2/ms [95% CI, 1.79–1.98 μm2/ms] for the complete group).
Figure 2 Time course of changes in the parameters used in diffusion tensor imaging (DTI) of the optic nerves of subjects during a 1-year period, by recovery group. Axial diffusivity is more altered on the baseline DTI scan for those with incomplete visual recovery (more ...)
Further statistical modeling revealed that time to presentation was not a significant predictor of recovery (P=.57), whereas AD continued to predict recovery status (P < .001). After subdividing the time of baseline imaging into early (0–15 days) vs later (16–31 days), we found that the early group displayed a mean AD of 1.51 μm2/ms (95% CI, 1.36–1.66 μm2/ms) when recovery was incomplete compared with 1.70 μm2/ms (95% CI, 1.60–1.80 μm2/ms) when recovery was complete. For those in the incomplete recovery group with later baseline imaging, the mean AD was 1.51 μm2/ms (95% CI, 1.37–1.66 μm2/ms), and for those in the complete group with later baseline imaging, the mean AD was 1.69 μm2/ms (95% CI, 1.48–1.90 μm2/ms). These results qualitatively suggest that predictive imaging can be achieved within 3 to 4 weeks of clinical onset.
RADIAL DIFFUSIVITY OVER THE NEXT 12 MONTHS
Radial diffusivity for the group was normal during the first month (divided into 0.91 μm2/ms [95% CI, 0.84–0.98 μm2/ms] at 2 weeks and 0.96 μm2/ms [95% CI, 0.88–1.03 μm2/ms] at 4 weeks) and progressively increased over the ensuing year (). The greatest rate of change was observed between months 1 and 2 (mean [SE], 0.085 [0.053] μm2/ms per month), followed by months 2 and 3 (mean [SE], 0.071 [0.054] μm2/ms per month). By 3 months, overall radial diffusivity was clearly elevated (1.11 μm2/ms [95% CI, 1.03–1.19 μm2/ms]) compared with baseline radial diffusivity, and it continued to remain elevated at 6 months (1.22 μm2/ms [95% CI, 1.13–1.30 μm2/ms]) and 12 months (1.30 μm2/ms [95% CI, 1.21–1.40 μm2/ms]). Mean diffusivity and fractional anisotropy, similar to radial diffusivity, became clearly abnormal by month 3, and they continued to be altered through month 12 ().
At 12 months, radial diffusivity in the incomplete recovery group (1.45 μm2/ms [95% CI, 1.31–1.59 μm2/ms]) could be differentiated from radial diffusivity in the recovered group (1.19 μm2/ms [95% CI, 1.10–1.28 μm2/ms]) (). Also, by 12 months, fractional anisotropy in the incomplete recovery group (0.25 [0.20–0.29]) could be differentiated from fractional anisotropy in the complete recovery group (0.33 [0.30–0.36]).