Abstract
Objectives: To assess the association between the thickness of the retinal nerve fiber layer (RNFL), assessed by optical coherence tomography (OCT), retinal periphlebitis (RP), and multiple sclerosis (MS) disease activity.
Methods: We studied a prospective cohort of 61 patients and 29 matched controls for 2 years, performing a neurologic assessment every 3 months and an ophthalmologic evaluation, including OCT scans, every 6 months. Baseline MRI studies were also carried out from which brain volume and lesion load were assessed.
Results: We found that the RNFL thickness in patients with MS was thinner than in controls, particularly in the temporal quadrant (p = 0.004). Although RNFL atrophy was greater in patients who also had optic neuritis (p = 0.002), it also augmented in MS patients who did not have optic neuritis compared with controls (p = 0.014). RNFL atrophy was correlated with greater disability (r = −0.348, p = 0.001) and longer disease duration (r = −0.301, p = 0.003). Furthermore, baseline temporal quadrant RNFL atrophy was associated with the presence of new relapses and changes in the Expanded Disability Status Scale by the end of the study (p < 0.05 in all cases). Indeed, RNFL thickness was correlated with white matter volume (r = 0.291, p = 0.005) and gray matter volume (r = 0.239, p = 0.021). The presence of RP was a risk factor for having new relapses in the next 2 years (odds ratio = 1.52, p = 0.02), and patients with RP had larger gadolinium-enhancing lesions volume (p = 0.003).
Conclusion: Retinal nerve fiber layer atrophy and the presence of retinal periphlebitis are associated with disease activity, suggesting that retinal evaluation can be used as biomarkers of multiple sclerosis activity.
The visual pathways are frequently affected by multiple sclerosis (MS),1 even in patients who have no visual disturbances.2,3 Moreover, retinal periphlebitis (RP) is an asymptomatic finding in 10 to 20% of patients.4–6 Optical coherence tomography (OCT) is a reliable technique for measuring the retina nerve fiber layer (RNFL) thickness.7,8 Previous studies reported the association between RNFL atrophy measured with OCT and visual impairment, abnormal visual evoked potentials, and optic nerve atrophy in patients with severe optic neuritis.9–12 In addition, the RNFL proved to be thinner in patients with MS.13 Because the retina is part of the CNS, it is commonly affected in MS and their axons lack myelin and because it is accessible for clinical examination, measuring RNFL thickness may be a useful biomarker14 to assess axonal loss in patients with MS.
In this study, we evaluated the association and diagnostic accuracy of RNFL thicknesses measured with OCT and the presence of RP for predicting disease activity in patients with MS. We studied a cohort of patients in the early to intermediate phase of their disease in the absence of a bias for patients with clinical involvement of the visual pathway. In addition, we assessed the relationship between RNFL atrophy and CNS atrophy and lesion load and between the presence of RP and CNS active lesions on MRI.
METHODS
Subjects.
We studied a cohort of 61 consecutive patients with MS (revised McDonald criteria15) in the early to middle phase of their disease as well as 29 sex- and age-matched healthy controls (HCs). Patients were recruited by their neurologist after obtaining informed consent. The Ethical Review Committee of the University of Navarra had approved the study. The inclusion criteria were having MS for <10 years, irrespective of the disease subtype or use of immunomodulatory drugs, and the ability to provide informed consent. We excluded patients with a disease duration >10 years with high disability (Expanded Disability Status Scale [EDSS] score >7.0) or with ophthalmologic diseases that might impair or bias OCT measurements (e.g., diabetes, glaucoma). Patients were included before performing any ophthalmologic or MRI examinations. Baseline neurologic and ophthalmologic examinations and OCT and MRI studies were performed in the same month.
Clinical assessment.
Neurologic evaluation was performed every 3 months for 2 years and included assessment of new relapses and disability and use of immunomodulatory therapy. We scored disability using the EDSS,16 MS Severity Scale,17 and MS Functional Composite (MSFC)18 by trained personnel. To define disability progression, we assessed the change in the EDSS score confirmed in a second visit after 6 months.19 Ophthalmologic evaluations were performed every 6 months for 2 years (five evaluations) by blinded ophthalmologists and included visual acuity using high-contrast Snellen carts; direct ophthalmoscopic examination after pupil dilation using 1% tropicamide to assess the presence of uveitis, RP, or vitreal cells; and the measurement of RNFL thickness by OCT. Visual impairment was defined as a visual acuity (VA) ≤0.8 (equivalent to 20/40) after refractory correction and normalized to contralateral eye visual acuity. The presence of RP was recorded as being focal or diffuse and arterial or venous and the number of RP for each eye and the quadrant location. Because retinal angiography did not increase the identification of RP at the baseline visit (data not shown), we did not repeat this examination in the following visits.
OCT.
Thickness of the RNFL was measured in each eye by an experienced technician, blind to the results of other studies, using OCT with StratusOCT 3000 and OCT 4.0 software (Carl Zeiss, Dublin) using the RNFL thickness 3.4 protocol with no pupil dilatation. Pupil dilatation has been shown to have little impact on OCT values and reproducibility,7 and its avoidance is preferred by patients. All scans obtained met the signal strength requirements of >7 (maximum 10) and involved uniform brightness across the scan circumference. We obtained the average and quadrant (temporal, superior, inferior, and nasal) RNFL thickness in micrometers. Using the normative database of age-matched control subjects provided by the OCT 4.0 software, we assigned a rank of normal (>5th percentile) or below normal (5th and 1st percentiles).
MRI studies.
MRI studies were performed using a 1.5-T Siemens Maestro Class Symphony (Erlagen, Germany). Three-dimensional T1-weighted scans with 88 contiguous axial slices (256 × 256 matrix, field of view [FOV] = 256 mm, flip angle 30 degrees, echo time [ET] 4.6 msec, in-plane voxel size 0.98 × 0.98 mm2) with a single doses of gadolinium (0.1 mmol/kg) and T2-weighted/proton density images with 48 contiguous axial slices (256 × 256 matrix, FOV = 256 mm, TE 120 msec, in-plane voxel size 0.47 × 0.47 mm2) were acquired from all patients. MRI studies were performed at the time of the first visit. No subjects were experiencing a clinical reactivation of the disease at the time that the scans were performed, and no patients were taking steroid treatment within the month prior to neuroimaging. We used MRIcro software (Chris Rorden, University of Nottingham, UK) to manually delimit the lesions in the T1- and T2-weighted scans of all patients (intraclass correlation coefficient = 0.892 [p < 0.001]). The volume of T1-, T2-, and gadolinium-enhancing lesion in each patient was calculated by multiplying the total number of lesion voxels by the size of the voxel. To quantify gray matter (GM) or white matter (WM) volume, we performed a voxel-based morphometry (VBM) analysis using the three-dimensional T1-weighted studies and the SPM2 software (Wellcome Department of Cognitive Neurology, University College of London, London, UK), running under Matlab v. 6.5 (Mathworks Inc., Natick, MA). We used a modified protocol of the optimized VBM method optimized for MS as described elsewhere20 to obtain the normalized and segmented images of each subject but avoiding the bias introduced by WM lesions. A trained neurologist blind to the results of the ophthalmologic examinations performed MRI analysis.
Statistical analysis.
Different statistical tests were used to analyze group differences or longitudinal changes depending on the normal distribution (t test and analysis of variance) or not (Mann-Whitney U test, Wilcoxon test, Kruskal-Wallis test, and Friedman test) of the variables. The normal distribution of all variables was assessed using the Kolmogorov-Smirnov test. The correlation between RNFL and clinical variables such as disability was carried out using the bivariate correlation (Pearson's or Spearman's correlation). Multiple linear regressions were used as a multivariate analysis to assess independent association between lesion load or whole GM/WM volumes and RNFL thickness (adjusted by sex and age). Logistic regressions were used to assess risk factors in the RP subanalysis. The level of significance was set at p < 0.05. For this analysis, we used the statistic package SPSS 13.0 (SPSS Inc., Chicago, IL). Because each subject has two RNFL measurements that might differ (one from each eye), we used the RNFL thickness from the eye showing the smaller RNFL thickness to calculate the predictive value of RNFL thickness. We calculated the diagnostic accuracy of the RNFL thickness using three different cutoffs points: the 1% and 5% percentiles of the normative data from the OCT 4.0 software database or an RNFL thickness <2 SDs of our HC group.
RESULTS
Clinical characteristics of the patient cohort.
Between January and April 2004, we recruited 61 patients with MS to this study. The patients were followed until April 2006. The baseline characteristics of the patients are shown in table 1. Patients display mild to moderate disability and medium disease duration. At the time of inclusion in the study, 29 patients were being treated with interferon beta and two with chemotherapy. None of the patients refused to undergo ophthalmic examinations at the time of entry into the study, including OCT scans. From the original 61 patients, three were lost for the clinical follow-up period. Of these, one was ruled out of the study immediately after inclusion because of the detection of active ischemic cardiopathy. The second of these patients relocated to another area in the middle of the follow-up and declined to continue in the study. Finally, the third patient refused to undergo direct ophthalmic examination, but she agrees to continue with the neurologic assessment. During the follow-up, the relapse rate was 1.23 (SD = 1.45) for the 2-year period. Accordingly, 26 patients did not experience relapses (relapse-free patients), 13 had one relapse, nine had two relapses, and 12 had three or more relapses. During the clinical follow-up, 14 patients started new immunomodulatory therapy. Of the clinically isolated syndrome (CIS) patients, 12 converted to relapsing-remitting MS (RRMS) because they experienced a second relapse and five RRMS patients converted to secondary progressive MS (SPMS). The change in the EDSS score during the 2-year period was 0.41 (SD = 0.78; range 0 to 3.5) and disability progression was confirmed in 12 patients at 6 months.
Table 1 Baseline demographics of the patients and HC group
Of the patients studied, 21 had experienced a previous episode of optic neuritis, 13 in the right eye, five in the left eye, and three bilaterally, enabling us to analyze 24 optic neuritis eyes. Thus, we analyzed 122 MS eyes: 24 optic neuritis eyes and 98 eyes without optic neuritis. Of the 24 optic neuritis eyes, 13 had no visual impairment, nine had mild visual impairment, and only two had a moderate to severe visual impairment. The mean VA of optic neuritis eyes was 0.83 (SD = 0.21; range 0.3 to 1.0). During the 2-year follow-up, five patients had novel unilateral optic neuritis (two cases at month 3, another two cases at month 9, and another two cases by month 15, including one patient who had two episodes of unilateral optic neuritis by months 3 and 15), leading to six new optic neuritis eyes. The presence of optic neuritis led to mild visual impairment (20/40) in only two cases (two eyes).
RNFL thickness and disease activity in patients with MS.
We found that the average RNFL thickness was less in patients with MS at study entry (mean RNFL thickness was 85.8 (SD = 13.9) μm in patients) vs HC (92.3, SD = 16.7) μm; p = 0.004). When analyzing the RNFL quadrants, we found that patients with MS have a reduction in the RFNL thickness in all quadrants except the nasal quadrant (figure 1A, p < 0.05 in all cases). Of the 61 patients, 11 (18%) were below the 1% percentile and 24 (39.3%) were below the 5% percentile of the normative database of the OCT 4.0 software at baseline.
Figure 1 Differences in the retinal nerve fiber layer (RNFL) thickness between patients with multiple sclerosis (MS) and controls
(A) Differences in the average, temporal, nasal, superior, and inferior quadrant RNFL thickness between patients (solid columns) and healthy controls (open columns) at baseline. Results are expressed as mean ± SD. *p< 0.05. (B) Longitudinal (months 0 [baseline], 6, 12, 18, and 24 [end of the study]) average RNFL thickness and temporal quadrant RNFL thickness in patients and healthy controls. Results are expressed as the mean RNFL thickness. *p< 0.05 at every time point. Patients with MS have a longitudinal decrease in the average and temporal quadrant RNFL thickness (p< 0.001 in both cases), and changes in the control group were also significant. **Significant longitudinal change from baseline to month 24.
When analyzing the effect of previous optic neuritis, we found that average and for each quadrant (except for nasal quadrant) RNFL thickness, optic neuritis eyes have a thinner RNFL than HC (p < 0.05 in all cases). Moreover, eyes without optic neuritis also display RNFL atrophy compared with HC (p < 0.05 in all cases). However, we found no differences in the RNFL thickness between optic neuritis eyes and eyes without optic neuritis, probably because our cohort was not selected from patients with permanent visual impairment after optic neuritis and because the number of patients with permanent visual impairment and the degree of the VA deficit were small. When comparing the disease subtype, we found that average and quadrant RNFL thickness (except for nasal quadrant) was diminished in RRMS patients compared with HC (p < 0.001 in all cases). Likewise, the RNFL was thinner in the temporal quadrant in CIS and progressive patients (SPMS and PPMS) compared with HC (p < 0.05 in all cases). Moreover, the baseline average RNFL thickness was correlated with disability at the time of the study entry (table 2).
Table 2 Correlation between RNFL thickness and disease activity
After 2 years, we found a decrease in the average RNFL thickness of 4.8 μm in patients (p = 0.01) and 2.2 μm in HCs (p = 0.03). Indeed, by the end of the study, the mean RNFL was 82.0 (SD = 15.6) μm in patients and 91.0 (SD = 12.1) μm in HCs (p = 0.004). Moreover, 23 of 58 patients (39.6%) were below the 1st and 5th percentiles of the normative data from the control population. The differences in the average and temporal quadrant RNFL thickness maintained between MS and controls at all time points (figure 1B). We identified six optic neuritis relapses in five patients, although none of these new optic neuritis led to a permanent visual deficit. RNFL thickness was not affected by the concurrent presence of optic neuritis because no changes were detected in RNFL thickness when comparing RNFL measures before and after the relapse (data not shown).
We found that patients with more active disease have a thinner temporal quadrant RNFL compared with stable patients. First, patients with new relapses during follow-up had a thinner RNFL in the temporal quadrant than relapse-free patients by the end of the study (p = 0.02). Additionally, the thickness of the RNFL in the temporal quadrant correlated with the number of relapses (table 2). Furthermore, patients with a confirmed progression at 6 months also had a thinner RNFL in the temporal quadrant than stable patients (p = 0.027). The temporal quadrant RNFL thickness correlated with the change in the EDSS score (table 2). We also analyzed whether the change in the average and in the quadrants RNFL thickness in the first 6 and 12 months of the study (from baseline to 6- and 12-month follow-up) was associated with changes in disease activity. We found that using the change in the RNFL thickness in the first 6 or 12 months was not more informative that the absolute value of the RNFL thickness at baseline (data not shown). Indeed, we found that patients treated with immunomodulatory drugs by the end of the study have a smaller decrease from month 0 to month 24 in both the average (p = 0.033) and temporal quadrant (p = 0.015) RNFL. We found that RNFL measurement has good specificity to identify patients who were going to have an increase in their EDSS score, although with low sensitivity (table 3). The test that offered the best result was the assessment of the temporal quadrant RNFL thickness using the 5th percentile cutoff of the normative database from the OCT 4.0 software (table 3).
Table 3 Diagnostic accuracy of the measurement of the RNFL thickness in identifying subjects whose disability will worsen over the next 2 years (disability progression) and of the occurrence of RP in identifying subjects who are going to experience new relapses in the next 2 years
Because MRI studies are the gold standard for surrogate outcomes for MS, we evaluated the association between changes in the RNFL thickness and lesion load or brain atrophy in the MRI. We found that average RNFL thickness at baseline correlated with baseline WM and GM volume (figure 2). We found no significant correlation with the volume of T1-, T2-, and gadolinium-enhancing lesions.
Figure 2 Correlation between average retinal nerve fiber layer (RNFL) thickness and gray matter (GM) and white matter (WM) volume on the MRI
Correlation between average retinal nerve fiber layer (RNFL) thickness and gray matter (GM) and white matter (WM) volume on the MRI. Average RNFL thickness at baseline correlated with WM volume (r = 0.291, p = 0.005) and GM volume (r = 0.239, p = 0.021).
RP and disease activity.
Of the patients studied, six (9.8%) had RP at baseline (one diffuse and five focal) that resolved completely or partially within 6 months. During the clinical follow-up, four new patients developed new RP, leading to a final RP accumulated frequency of 10 of 61 patients (16%). Interestingly, the occurrence of relapses in the previous year predicted the presence of RP (odds ratio [OR] = 1.13; 95% CI: 1.02 to 1.26; p = 0.004). Moreover, patients with RP have more relapses in the follow-up period (p = 0.005), and indeed, the existence of RP was a risk factor for having new relapses (OR = 1.52; 95% CI: 1.36 to 1.64, p = 0.002) but not for disability progression. When we studied the diagnostic accuracy of RP in predicting relapses, we found that the occurrence of RP had moderate accuracy (71%) in identifying patients who are going to experience more relapses (more than three relapses in the next 2 years, table 3). However, we found no association between the occurrence of RP and previous optic neuritis or with the use of immunomodulatory therapy. Finally, patients with RP at baseline had a larger gadolinium-enhancing lesion volume than patients without RP (p = 0.003).
Finally, we evaluated whether RP induce local tissue damage by analyzing its effect on the regional thickness of the RNFL. We assessed the association between the occurrence of RP in a given quadrant (temporal, nasal, superior, or inferior) and the corresponding RNFL thickness in that quadrant 6 months before, during, and 6 months after the resolution of the RP.12 We found that the occurrence of RP in a given quadrant was not associated with a significant thinning of the RNFL in the same quadrant (data not shown).
DISCUSSION
In this study, we evaluated the utility of assessing retinal abnormalities such as RNFL thickness and the presence of RP as biomarkers for MS. The underlying hypothesis was that RNFL thickness might be informative of the widespread axonal damage of the CNS associated with MS and that RP might reflect CNS inflammatory activity. We found that RNFL atrophy is an early and frequent phenomenon in MS, even in patients who have not suffered clinical involvement of the optic pathway.13 Thus, our findings are in agreement with previous studies showing that axonal damage is an early phenomenon in MS.1,21 In our study, we found no differences in the RNFL thickness between optic neuritis eyes and eyes without optic neuritis as described previously.7,10–12 This probably reflects the fact that our cohort was not selected from patients with permanent visual impairment, and only two patients (two eyes) had a moderate to severe visual impairment after optic neuritis. Thus, we might not have enough power to confirm such results. Indeed, we used only a high-contrast visual test and not the recently validated low-contrast visual test (Sloan charts),22,23 which have higher sensitivity to identify changes in patients with MS and might allow to improve our ability to identify correlations with the RNFL thickness. We believe that the association of RNFL atrophy with new relapses is secondary to the fact that patients with more relapses tends to have more active disease, which in turn leads to more tissue damage. Histologic studies revealed that an RFNL loss of 5,000 axons per year24,25 and an RNFL thickness decrease of 1.7% per decade.26 In our study, we found a slight decrease in the RNFL thickness in the control group after 2 years. However, such a finding is in the resolution limit of the OCT and requires further validation in larger cohorts. During the follow-up period of our study, we found that the RNFL thickness in patients with MS decreased more than would normally be expected with age, indicating that axonal damage is an ongoing process in patients with MS.
We found a moderate correlation between GM and WM volume in the MRI and RFNL thickness, suggesting that both techniques measure a similar phenomenon such as axonal loss.27 The fact that the correlations were only moderate might be explained by the fact that RNFL thickness is a local measurement of the global brain atrophy. However, even if thickness of the RNFL might be less informative compared with global MRI measurements, its simplicity, low cost, and patient convenience might make this measure a convenient test for frequent assessment of disease evolution or even to monitor response to therapy. We hypothesize that frequent prospective RNFL studies might improve the sensitivity of this technique.
Although the pathogenesis of RP is unknown, RP might be the retinal counterpart of the inflammatory process affecting the CNS. However, RP affects an area devoid of myelin, which has to date been considered the main target of the autoimmune response.28 Thus, the presence of RP raises many questions about the disease pathogenesis. In this study, we found that RP is associated with a more active disease in terms of relapses and presence of gadolinium-enhancing lesions. The association between RP and clinical and MRI markers of the inflammatory process27 strongly suggests that RP is not a phenomenon that is independent of the CNS inflammation, but might have the same pathogenic origin as the other perivenular lesions in the CNS. Some authors have suggested that the initial target of the inflammatory process might be different of myelin sheaths and might represent an immunologic attack on components of the blood-brain barrier (Prof. Martin Raff, personal communication).
When considering the usefulness of RP as a biomarker for MS, we found that the presence of RP identifies individuals at high risk of experiencing active disease in terms of relapse rate. The main limitation is that RP prevalence might not be high enough (10 to 20%) to become a useful screening for patients with MS. Moreover, in this study, we did not have enough power to assess whether RP is influenced by immunomodulatory drugs. Considering the hypothesis that the pathogenesis of RP is related to the pathogenesis of CNS inflammatory infiltrates and that the latter is the main target of immunomodulatory drugs, we can speculate that RP might be useful to monitor response to therapy or to identify nonresponders to immunomodulatory therapy. Depending on the different criteria, the percentage of nonresponders range from 20 to 40% of patients.29,30 Thus, future studies will be required to establish whether RP is useful to monitor response to immunomodulatory therapy.
ACKNOWLEDGMENT
The authors thank Aurora Alvarez-Vidal for help performing OCT measurements, Drs. Xavier Montalban and Mar Tintore from the Hospital Vall d'Hebron, Barcelona, for helpful comments, and Prof. Martin Raff from the University College London, for comments on the role of retina inflammation in the pathogenesis of MS. They also thank the Navarra MS Society and the patients for their participation.
Footnotes
-
*Both authors contributed equally to this work.
Supported in part by the Spanish Ministry of Health (FIS PI051201), the “Fundacion Uriach” and by an unrestricted grant by Gemac SA (Cenon, France) to P.V. J.S. was a fellow of the Spanish Ministry of Health (FIS CM #05/00222).
Disclosure: The authors report no conflicts of interest.
Received September 16, 2006. Accepted in final form December 31, 2006.
REFERENCES
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Topics Discussed
Alert Me
Recommended articles
Optical coherence tomography and disease subtype in multiple sclerosis
No evidence of disease activity is associated with reduced rate of axonal retinal atrophy in MS
Sex-specific differences in retinal nerve fiber layer thinning after acute optic neuritis
Optical coherence tomography is highly sensitive in detecting prior optic neuritis