MRI signal intensity varies along the course of the normal optic nerve

The optic nerve consists of a bundle of retinal ganglion cell fibers that extends from the globe to the optic chiasm in the brain and can be divided into four segments: intraocular, intraorbital, intracanalicular, and intracranial^1,2. Optic neuropathies may result from a wide variety of causes, with different workups and treatment plans dependent upon the etiology. Optic nerve imaging is an important diagnostic tool that offers insights beyond laboratory and historical findings¹.

Magnetic resonance imaging (MRI) is routinely used to evaluate the health of the central nervous system and is particularly useful in differentiating underlying causes of abnormalities. With an optic neuropathy, clinicians rely on differentiating patterns of enhancement, which may indicate disease processes such as inflammation or swelling. Optic neuritis is an inflammatory condition of the optic nerve that is a frequent cause of vision loss in adults^3,4. There are various etiologies of optic neuritis that may present with different patterns of MRI enhancement. For instance, bilateral optic nerve enhancement and prechiasmatic optic nerve enhancement are strong predictors of neuromyelitis optica spectrum disorder (NMOSD)^5–9.

However, one limitation of MRI interpretation is the dependence on a qualitative assessment by the reviewer. Natural variations within healthy anatomical structures could pose a challenge to accurate interpretation of imaging, as normal variations may be misinterpreted as a pathologic process. Thus, identifying predictable patterns of signal variation in the optic nerve on MRI and quantifying such changes may prove helpful for more accurate interpretation. In this study, we sought to determine and quantify if the signal intensity of the normal optic nerve on MRI varies predictably from anterior to posterior positions on precontrast and postcontrast images.

Methods

Patients

Institutional review board approval was obtained for this Health Insurance Portability and Accountability Act–compliant retrospective study with a waiver of informed consent. Data were obtained from patients who underwent neuro-ophthalmic exam at the University of Minnesota Medical Center from January 2005 through April 2021. Inclusion criteria for patients were age 18 years or older at the time of the MRI, visual acuity of 20/25 or better, and no evidence of optic neuropathy on comprehensive neuro-ophthalmologic examination. Each patient underwent imaging for a cranial nerve 3, 4, or 6 palsy. Images were obtained at various hospitals and outpatient imaging centers and ported to our institution.

MRI Analysis

Following identification of optic nerves from patients who met criteria, a neuro-radiologist (MG) with over 8 years of experience performed the measurements of optic nerve intensity. These analyses were performed on 1) axial non-fat saturated pre-contrast T1-weighted and 2) axial and 3) coronal post-contrast, fat-saturated T1-weighted images. A total of 55 scans were 1.5 Tesla (T) images, while the remaining 20 were 3.0 T. Two locations along the optic nerve were selected in order to compare signal intensity: a mid-orbit optic nerve (MO-ON) segment and a pre-chiasmatic optic nerve (PC-ON) segment. The MO-ON segment measurement was performed 1.5 cm posterior to the posterior margin of the globe, while the PC-ON segment measurement was assessed 1 cm anterior to the optic chiasm. At each location, the signal intensity was measured on the PACS (Philips IntelliSpace 4.4, Amsterdam, Netherlands) program by selecting the smallest area that encompassed the entire cross-section of the optic nerve in that location. A reference signal intensity measurement was performed at normal-appearing temporalis muscle and a ratio of optic nerve intensity to reference intensity was calculated (Figures 1 and 2).

Figure 1.

Precontrast axial T1-Weighted images demonstrates intensity measurement from mid right optic nerve and temporalis muscle as reference (a). Region of interest (ROI) selection from pre-chiasmatic right optic nerve (b).

Figure 2.

Fat-saturated postcontrast axial T1-Weighted images demonstrates intensity measurement from mid right optic nerve and temporalis muscle as reference (a). ROI selection from pre-chiasmatic right optic nerve (b).

Statistical Analysis

The pre- and post-contrast measurements of PC-ON and MO-ON were summarized by the mean, standard deviation (SD), median and range for each eye. These two optic nerve locations were compared with a two-sided paired t-test. The percent differences (PC-ON – MO-ON)/MO-ON were reported as the mean, SD and 95% confidence interval (CI). This percent change was compared between pre- and post-contrast also with a paired t-test after averaging over the right and left eyes. The Pearson correlation coefficient measured the association between right and left eyes for the post-contrast PC-ON and MO-ON differences. To account for the correlation between the two eyes of the same person, a linear mixed model was performed to test for differences between the eyes while treating the eyes as a random effect. This statistical model allows for missing observations and the inclusion of other covariates of interest. All statistical analysis was carried out with SAS 9.4 (SAS institute Inc., Cary NC). P-values less than 0.05 were considered statistically significant.

Results

The initial database query yielded 750 patients with a cranial mononeuropathy, of which, 75 (41.3% women, mean age, 55.68 years) met the inclusion criteria. From these 75 individuals, 8 right and 7 left optic nerves were unable to be properly measured in the axial plane due inadequate views, yielding 67 right and 68 left optic nerves in total that were measured axially. Additionally, from the original 75 individuals, 11 right and 11 left optic nerves were unable to be adequately measured in the coronal plane due to absence of fat suppression, resulting in 64 right and 64 left optic nerves being measured coronally. The optic nerves were independently excluded from the original 75 individuals to form each group and thus the optic nerves excluded were not necessarily the same between groups. Across all 75 individuals, a total of 84% of the patients were White, 4% Black, and 5.3% Asian. Of the MRI scans, 13 (17.3%) demonstrated abnormalities, including intracranial masses or vascular changes, including hemorrhage and aneurysms, while the rest were normal. Of the pre-contrast T1 axial images, only 2 were performed with fat suppression.

Signal intensity ratios of the optic nerve was significantly higher in the PC-ON segment compared to the MO-ON segment on both pre-contrast and post-contrast T1-Weighted images (Tables 1 and 2). This percent difference was slightly higher on the pre-gadolinium images (mean 19.6% brighter) compared to the post-gadolinium images (mean 14.2% brighter), although it was not significant (p=0.257). Further, the mean signal intensity ratios were drastically lower on pre-contrast images in both PC-ON (Means: Pre-Contrast = 0.012, Post-Contrast = 0.953) and MO-ON (Means: Pre-Contrast = 0.011, Post-Contrast = 0.843) segments.

Table 1. Pre-chiasm optic nerve signal is significantly brighter than mid-orbital optic nerve.

Summary statistics for Right (R, N=67) and Left (L, N=68) optic nerve (ON) and the difference between ratios of midorbital (MO) and chiasm (PC) on pre-and post- gadolinium (gad) images to the reference (ref) intensity on axial and coronal (cor) slices.

Variable	Mean (SD)	Median [Min/Max]	PCON-MOON Mean (SD)	P-value
Axial R Pre-GAD MO-ON/Ref	1.01 (0.26)	1.01 [0.49/1.71]	0.153 (0.325)	<0.001
Axial R Pre-GAD PC-ON/Ref	1.16 (0.24)	1.12 [0.72/2.18]
Axial L Pre-GAD MO-ON/Ref	1.05 (0.28)	1.03 [0.47/2.07]	0.123 (0.330)	0.003
Axial L Pre-GAD PC-ON/Ref	1.18 (0.24)	1.13 [0.77/2.00]
Axial R Post-GAD MO-ON/Ref	0.84 (0.17)	0.84 [0.51/1.26]	0.095 (0.187)	<0.001
Axial R Post-GAD PC-ON/Ref	0.93 (0.19)	0.91 [0.59/1.38]
Axial L Post-GAD MO-ON/Ref	0.85 (0.15)	0.82 [0.59/1.28]	0.126 (0.174)	<0.001
Axial L Post-GAD PC-ON/Ref	0.98 (0.21)	0.96 [0.57/1.59]
Cor R Post-GAD MOON	1.04 (0.18)	1.01 [0.70/1.45]	0.128 (0.158)	<0.001
Cor R Post-GAD PCON	1.16 (0.23)	1.17 [0.67/1.62]
Cor L Post-GAD MOON	1.02 (0.18)	1.00 [0.65/1.47]	0.126 (0.133)	<0.001
Cor L Post-GAD PCON	1.15 (0.20)	1.15 [0.75/1.55]

Table 2. Percentage differences between pre-chiasmatic (PC) vs. mid-orbital (MO) optic nerve (ON).

Pre-chiasm optic nerve demonstrates significant brighter signal intensity compared to mid-orbital optic nerve on pre- and post-gadolinium (gad) axial and coronal (cor) images.

PC-ON/MO-ON	Mean (SD)	95% Conf Interval
Axial R Pre-GAD	22.6% (44.8)	11.8%, 33.4%
Axial L Pre-GAD	18.4% (38.5)	9.1%, 27.6%
Axial R Post-GAD	13.8% (23.6)	8.0%, 19.5%
Axial L Post-GAD	15.9% (20.9)	10.8%, 21.0%
Cor R Post-GAD	12.8% (16.2)	8.8%, 16.9%
Cor L Post-GAD	13.0% (14.1)	9.4%, 16.5%

There was a moderate correlation (r=0.45, p<0.001) in the PC-ON minus MO-ON difference between the right post-GAD and the left post-GAD (N=61). A linear mixed regression model (N=68 MRIs) including the right and left eyes as correlated observations, found that the estimated difference in signal intensity between the two eyes for the post-GAD PC-ON minus MO-ON difference was 0.031 (p=0.192) with the left eye having a larger, but non-significant, difference in brightness between PC-ON and MO-ON. When other factors were included in the model, age and gender were not significant with p-values 0.284 and 0.536, respectively. Cranial nerve diagnosis was significant (p=0.041) with brighter optic nerve signal intensity for patients with a 3^rd nerve palsy compared to patients with a 4^th or 6^th nerve palsy.

Notably, not all nerves showed higher intensity posteriorly. On pre-contrast images, 19 (25.3%) of the left optic nerves and 17 (22.7%) of the right optic nerves demonstrated less intensity posteriorly compared to the mid-orbit segment. Similarly, on post-contrast images, 16 (21.3%) left nerves and 18 (24%) right nerves demonstrated less brightness posteriorly.

Looking at differences between 1.5T and 3.0T MRI, on pre-contrast images, the 3T images showed a greater average difference (p<0.001) in PC-ON to MO-ON ratios. There was no significant difference seen on post-contrast images between machines.

Discussion

The results demonstrate significantly brighter mean intensity ratios of the pre-chiasmatic optic nerve compared to the mid-orbit optic nerve among patients without optic neuropathy. This subtle finding may have important implications for clinicians utilizing MR imaging for diagnostic purposes.

Clinicians typically employ a qualitative assessment of neuroimaging to prioritize the differential diagnosis. Patterns of optic nerve enhancement can help differentiate idiopathic demyelinating optic neuritis (IDON) from optic neuritis associated with NMOSD and myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD). Previous studies have described more posterior enhancement and bilateral enhancement of the anterior visual pathways in NMOSD as compared with IDON or MOGAD. The results of our study, demonstrating a propensity for the normal PC-ON to appear brighter on MRI than the normal MO-ON, indicate that care must be taken in labeling regions of enhancement. For instance, in the assessment of a patient with possible bilateral atypical optic neuritis, the clinician could reasonably mistake normal, post-contrast pre-chiasmatic brightness for subtle bilateral enhancement. These results imply that when suspicious for subtle PC-ON enhancement, the clinician should consider performing quantitative assessment of brightness differences and using a 20% signal difference threshold to account for normal signal variation.

Why would the PC-ON appear brighter on both pre-contrast and post-contrast images? It is possible that anatomical variations of the optic nerve between the mid-orbit and intracranial portions could account for the measured differences in signal intensities. We also hypothesize that the variability in surrounding structures (e.g., orbital fat vs CSF/bony skull base) may influence the measured signal differentially from the orbital to the cisternal optic nerve segments. There may be a difference in volume averaging, slice-to-slice signal variation, or B1 radiofrequency heterogeneity between these two locations as well. It is also possible that optic nerve movement may contribute to this finding. The MOON can move during image acquisition while the PCON likely will not. Further research into the anatomy of the optic nerve along its length may help explain the differences observed on MRI. Future studies could expand the normal percent signal differences and include additional points of measure along the optic nerve. They could also measure signal intensity of the PC-ON and MO-ON among eyes with optic nerve enhancement.

There are several limitations in this study. There are inherent limitations of all retrospective studies including ascertainment and referral bias and nonstandard (e.g., varying slice thicknesses and image resolutions) collection of MRI images. Images were obtained at different institutions on different machines. However, this heterogeneity makes it unlikely that our findings are the result of artifact from a single manufacturer. Asymmetric and irregular fat saturation could have affected our results. However, the consistent findings on the T1 pre-contrast non fat-saturated images argue against this concern. The same inconsistencies in MRI machines and imaging protocols that introduce variability into our study also potentially make our results more generalizable and relevant to clinical care. While most nerves in each group appeared brighter posteriorly as compared to a mid-orbital segment, there was a significant minority of nerves that demonstrated the opposite finding. All of our patients had ocular motor cranial nerve palsies, and while none showed any evidence of an optic neuropathy, it is unclear if the cranial nerve palsy could have affected the signal intensity of the optic nerve in a differential fashion. In fact, for unclear reasons patients with CN3 showed a statistically greater difference than those with a CN4 or CN6 palsy. Despite these limitations, we believe that our findings are valid.

In conclusion, we report that on average healthy optic nerves demonstrate significantly brighter intensity ratios on MRI imaging posteriorly, compared to mid-orbital segments on both pre- and post-contrast MRI. When faced with possible subtle enhancement of the prechiasmatic optic nerve, it seems reasonable to pursue quantitative intensity measurements along the optic nerve and determine if more than a 20% difference in brightness exists to suggest true optic nerve enhancement.