Correlation of MRI Findings With Patterns of Visual Field Loss in Patients With Pituitary Tumors

Background: Compression of the optic chiasm by pituitary tumors typically results in bitemporal hemianopia, implying that nasal retinal fibers are preferentially damaged. The reason for this is not clear. One theory suggests that nasal fibers are selectively vulnerable simply because they cross each other. This study investigated the "crossing theory" by correlating visual field (VF) loss with chiasmal elevation and with the degree of eccentric compression on MRI scans. Methods: Our hospital database was searched to identify patients with a) chiasmal compression by a pituitary tumor; b) documented preoperative evidence of VF loss; and c) preoperative MRI scan performed within 1 month of VF testing. Temporality and bitemporality indices were derived from pattern deviation VF plots. Elevations of the central and peripheral parts of the chiasm were obtained from MRI scans, from which the eccentricity of compression was calculated. Temporality indices and hemifield loss were compared with central chiasmal elevation, and nasal hemifield loss in each eye was plotted against eccentricity. Results: Eleven patients were suitable for analysis. The degree of bitemporal VF involvement was significantly correlated with elevation of the central chiasm (P = 0.004). However, there was minimal involvement of nasal VFs, and no demonstrable increase in nasal field loss with increasing eccentricity of compression. Conclusions: This study provides support for the crossing theory. These findings will inform further finite element models of chiasmal compression. A larger, prospective study is planned.

Original Contribution Correlation of MRI Findings With Patterns of Visual Field Loss in Patients With Pituitary Tumors Emily J. Kane, BMedSci, David E. Ashton, MBBS, FRANZCR, Peter J. Mews, MBBS, FRACS, Kate Reid, MBBS, FRANZCO, Andrew Neely, BEng, MEngSc, PhD, FRAeS, Christian J. Lueck, MB BChir, MA, PhD, FRACP FRCP(UK), FAAN Background: Compression of the optic chiasm by pituitary tumors typically results in bitemporal hemianopia, implying that nasal retinal fibers are preferentially damaged. The reason for this is not clear. One theory suggests that nasal fibers are selectively vulnerable simply because they cross each other. This study investigated the "crossing theory" by correlating visual field (VF) loss with chiasmal elevation and with the degree of eccentric compression on MRI scans. Methods: Our hospital database was searched to identify patients with a) chiasmal compression by a pituitary tumor; b) documented preoperative evidence of VF loss; and c) preoperative MRI scan performed within 1 month of VF testing. Temporality and bitemporality indices were derived from pattern deviation VF plots. Elevations of the central and peripheral parts of the chiasm were obtained from MRI scans, from which the eccentricity of compression was calculated. Temporality indices and hemifield loss were compared with central chiasmal elevation, and nasal hemifield loss in each eye was plotted against eccentricity. Results: Eleven patients were suitable for analysis. The degree of bitemporal VF involvement was significantly correlated with elevation of the central chiasm (P = 0.004). However, there was minimal involvement of nasal VFs, and no demonstrable increase in nasal field loss with increasing eccentricity of compression. Conclusions: This study provides support for the crossing theory. These findings will inform further finite element models of chiasmal compression. A larger, prospective study is planned. Journal of Neuro-Ophthalmology 2019;39:333-338 doi: 10.1097/WNO.0000000000000763 © 2019 by North American Neuro-Ophthalmology Society Australian National University Medical School (EJK, PJM, KR, CJL), Canberra, Australia; Departments of Neurology (EJK, CJL), Radiology (DEA), Neurosurgery; (PJM), and Ophthalmology (KR), The Canberra Hospital, Canberra, Australia; and School of Engineering and Information Technology (AN), UNSW, Canberra, Australia. The authors report no conflicts of interest. Address correspondence to Emily J. Kane, BMedSci, Australian National University Medical School, 54 Mills Road, Acton 2601, Australia; E-mail: [e-mail] Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 T he optic chiasm contains nerve fibers from the optic nerves that partially decussate before forming the optic tracts. The nasal retinal fibers from each eye decussate at the chiasm to join the uncrossed temporal retinal fibers from the other eye in the contralateral optic tract. Topographic representation of the visual fields (VFs) is very precisely preserved in the chiasm (1). Compression of the optic chiasm by pituitary tumors generally results in selective loss of the temporal VFs, or bitemporal hemianopia (2), implying that the nasal retinal fibers are preferentially damaged. The reason for this preferential damage is not fully understood. One explanation is the "anatomical theory," in which nasal fibers are selectively vulnerable because of their anatomical location in the center of the chiasm. Various authors have suggested that the basis for this anatomical theory is that the center of the chiasm, containing the nasal fibers, is subject to the greatest pressure (3) and/or stretch (4) as a result of extrinsic compression. It also has been proposed that nasal fibers in the center of the chiasm may be more susceptible to compression-induced ischemia due to the anatomy of the blood supply (5). Any or all of these factors may contribute to the increased susceptibility of nasal fibers to chiasmal compression. However, they should also affect the temporal fibers and hence the nasal VFs, albeit to a lesser extent. Thus, a purely anatomical theory does not, by itself, account for the highly selective damage to nasal fibers that manifests as bitemporal hemianopia often with complete sparing of the nasal fields. An alternative explanation that could account for the selective damage to nasal fibers is the crossing theory. McIIwaine et al (6) hypothesized that nasal fibers are selectively vulnerable simply because they cross each other: crossing fibers have a smaller area of contact than fibers that run parallel to each other. Therefore, any compressive force applied to the chiasm will result in greater stress on fibers 333 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution which cross, compared to those that do not. If this crossing theory is correct, nasal fibers should be particularly vulnerable to compression independent of the anatomical location of chiasmal compression. Specifically, wherever the chiasm is damaged, there should be proportionately greater involvement of the temporal fields. The crossing theory has previously been investigated using finite element modeling (7,8), but the model remains limited by a lack of precise anatomical information. Because of the precise topographical arrangement of nerve fibers in the chiasm, the pattern of VF loss arising from chiasmal compression should be predictable. Previous studies (9) have not examined the relationship between VF loss and precise location of chiasmal compression in sufficient detail to test the crossing theory. The aim of our study was to test the crossing theory by comparing observed patterns of VF loss with what would be predicted from the location of chiasmal compression. For example, lesions affecting the center of the chiasm should (and do) produce a typical bitemporal hemianopia (1,2,10). However, involvement of the far lateral aspect of the chiasm should damage uncrossed temporal fibers, resulting predominantly in ipsilateral nasal VF loss. Clinical experience suggests this is relatively uncommon, whereas a binasal hemianopia (which would be predicted to result from extrinsic, constrictive compression of the chiasm) is extremely rare (1,2). This study set out to test 3 hypotheses: 1. Increasing elevation of the optic chiasm should result in increasing bitemporal VF loss (this would hold true for both anatomical and crossing theories). 2. If the anatomical theory is correct, increasing elevation of the chiasm should begin to cause an increasing amount of nasal VF loss, albeit less than the degree of temporal VF loss. Conversely, the crossing theory would suggest that the nasal VFs will continue to be relatively spared as chiasmal elevation increases. 3. Similarly, if the anatomical theory is correct, increasing eccentricity of chiasmal compression should result in increasing ipsilateral nasal VF loss. Conversely, the crossing theory would predict less involvement of the nasal fields in this situation. METHODS Patient Selection To be eligible for our study, patients had to have a) chiasmal compression by a pituitary tumor, b) documented preoperative evidence of VF loss, and c) a preoperative MRI scan performed within 1 month of VF testing. Potential cases of pituitary compression with VF loss were identified from records of the Departments of Ophthalmology and Neurosurgery at The Canberra Hospital between 2009 and 2016. VFs and MRI scans were deidentified at source and analyzed off-line. This study was approved by the ACT Health Human Research Ethics Committee (ETHLR.16.122) and the Australian National University Human Research Ethics Committee (protocol 2016/625). Data Analysis: Visual Fields For each patient, every point within each quadrant of the pattern deviation plot was assigned a value from 0 to 4 (Fig. 1). Each quadrant of the pattern deviation plot was assigned a score based on an arbitrary assignment of the statistical scores (Table 1). Pattern deviation plots (as opposed to the gray scale or total deviation plots) were used to highlight localized and severe defects (such as those related to lesions of the visual pathway) and minimize the effects of generalized VF loss (e.g., from cataracts). All included VF tests had false-positive rates of less than 10%. Two patients had falsenegative rates greater than 15% but were included in the study because we believed that false responses due to inattention would probably be distributed equally across temporal and nasal VFs and so would be corrected for in our calculations of the bitemporality index. Once the pattern deviation score was obtained, the total for each quadrant was calculated and then divided by the maximum possible score to yield a normalized score (Table 1). A higher normalized score represented a more profound VF deficit. A temporality index was calculated as the difference between the sum of the normalized temporal quadrant scores and the sum of the normalized nasal quadrant scores, multiplied by 100 (Table 1). The temporality index could range from 2100 (complete nasal VF loss) to +100 FIG. 1. Scoring system of pattern deviation plots in patients with chiasmal compassion. 334 Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 1. Calculation of temporality and bitemporality indices Scores calculated for each quadrant of the VFs in Figure 1 Left VF Quadrant Upper Lower Right VF Temporal (max. 72) Nasal (max. 76) Nasal (max. 76) Temporal (max. 72) 69 72 10 59 19 30 69 72 of normalized nasal quadrant scores Temporality index: Temporality index ¼ ðSum of normalized temporal quadrant scores2sum Þ · 100 2 ð69þ72Þ2ð 19 þ 30 Þ For the right VF above, the temporality index would be as follows: Temporality index ¼ ð 72 72 2 76 76 Þ · 100z þ 66 The right VF has a predominantly temporal VF defect. The corresponding left VF has a similar pattern of VF loss but with a lower temporality index because of the greater nasal VF involvement (temporality index z +53). The bitemporality index was calculated as the average of the 2 temporality indices for the left and right eyes' VFs. In the above example, the bitemporality index was approximately +59. VF, visual field. (complete temporal VF loss). A temporality index of zero indicated that nasal and temporal VFs were equally affected. A bitemporality index was then calculated as the average of the temporality indices of the 2 eyes. The bitemporality index also could range from 2100 (complete nasal VF loss in both eyes, no temporal loss in either eye) to +100 (pure bitemporal hemianopia). Data Analysis: MRI Scans Chiasmal Elevation For each patient, 3 adjacent coronal slices that included the optic chiasm were identified on the MRI scans, which had been performed with standard 5-mm interslice spacing. The optic canals were identified on the most anterior section, and the junctions of the optic tracts with the brain were identified on the most posterior section. It was assumed that these points were relatively fixed, and that the normal chiasm would lie on a plane bounded by these 4 points. Using Syngo imaging Software (Siemens Healthineers, Erlangen, Germany), 2 lines joining ipsilateral optic canals and optic tracts were drawn, and the predicted position of the chiasm could then be mapped onto the middle segment (Fig. 2A). The elevation of the chiasm was measured as the vertical displacement from the expected position at 3 points: the leftmost border, the center, and the rightmost border of the chiasm. Chiasmal Eccentricity To qualify for eccentric compression, patients had to demonstrate a clear difference between right- and leftsided chiasmal elevation. Subjects with equal elevation on the right and left were excluded. The "eccentricity" of chiasmal compression was calculated by subtracting the elevation of the center of the chiasm from the elevation of the higher side (either right or left) of the chiasm in the coronal plane. Positive eccentricity indicated that one side of the chiasm was elevated more than the center (Fig. 2B). Analysis of the VFs and MRI scans were performed by 2 independent observers; the individual who scored the VFs was blinded to the MRI measurements, and vice versa. Regression statistics were calculated using SPSS. Hypothesis Testing Hypothesis 1 The relationship between chiasmal elevation and bitemporal VF loss was determined for each subject by comparing the FIG. 2. A. Measurement of the vertical displacement (orange arrows) of the optic chiasm (light blue line) from its expected position (dark blue line) on the coronal MRI scan including the center of the chiasm. The 2 red dots represent points on the 2 lines drawn from the optic canals to the ipsilateral optic tracts (see text); T, pituitary tumor. B. Examples of positive and negative eccentricity are shown. Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 335 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution FIG. 3. A. Relationship between bitemporality index and elevation of the center of the chiasm for individual subjects. Gray lines represent the 95% confidence intervals. B. Relationship between temporal (blue) and nasal (orange) VF loss and chiasmal elevation for individual eyes. Predicted relationships based on an anatomical model and the crossing theory are illustrated at the right of the figure. The anatomical model suggests that nasal field involvement should increase proportionately to chiasmal elevation, although not as much as temporal field involvement. Crossing theory predicts relatively little nasal field involvement because the parallel (uncrossed) temporal fibers should be relatively protected. C. Relationship between nasal hemifield loss and eccentricity of chiasmal compression for ipsilateral (red) and contralateral (green) eyes. Predicted relationships based on an anatomical model and crossing theory are illustrated at the right of the figure. The anatomical model suggests that ipsilateral nasal field involvement should increase with increasing eccentricity, whereas there should be no, or decreasing, involvement of the contralateral eye. The crossing theory suggests that neither nasal field would be particularly affected by compression, but that there might be a slight tendency toward increasing involvement of the ipsilateral eye. bitemporality index with the degree of elevation of the center of the chiasm (in mm). Hypothesis 2 For each eye, the nasal and temporal VF scores were compared with the degree of elevation of the center of the chiasm (in mm). Hypothesis 3 The ipsilateral and contralateral nasal VF loss scores were compared with the degree of eccentricity of compression (in mm), as defined above. RESULTS Patients One hundred thirty-one patients initially were identified with suspected pituitary tumors. Of these, only 11 had both 336 preoperative VFs and an acceptable-quality MRI scan performed less than 1 month apart. Because of tumor recurrence, 2 patients were assessed and scanned on 2 occasions, yielding a total of 13 VF-MRI pairs. Effect of Chiasmal Elevation Hypothesis 1 Increasing elevation of the center of the chiasm was significantly associated with an increase in bitemporality index (P = 0.004) (Fig. 3A). The relationship remained significant (P = 0.006) when reanalyzed without the second set of measurements for the 2 patients with 2 VF-MRI pairs. Hypothesis 2 Looking at all eyes individually, increasing elevation of the chiasm was associated with increasing loss of the temporal Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution VFs (P = 0.053) (Fig. 3B). There was no suggestion of a significant relationship between increasing chiasmal elevation and nasal VF loss (P = 0.58). Hypothesis 3 All subjects demonstrated negative scores for eccentricity, that is, the middle of the chiasm was elevated more than the periphery. Six subjects demonstrated a clear difference between right- and left-sided elevation and were considered to have eccentric compression. The relationships between nasal hemifield loss for the ipsilateral and contralateral eyes (relative to the higher side of the chiasm) and eccentricity are illustrated in Figure 3C. A weak negative association between nasal hemifield loss in the contralateral eye and eccentricity was observed, although this failed to reach significance (P = 0.32). However, there was no significant relationship between nasal hemifield loss in the ipsilateral eye and eccentricity (P = 0.86). DISCUSSION Our study demonstrated a significant positive relationship between the extent of bitemporal VF loss and the degree of central chiasmal elevation (Fig. 3A). Looking at individual eyes, increasing chiasmal elevation was associated with increased temporal VF loss, but the nasal VFs were relatively unaffected (Fig. 3B). Although there was a suggestion of reduced contralateral nasal field loss with increasing eccentricity, there was no evidence of a relationship between eccentricity and ipsilateral nasal hemifield loss (Fig. 3B). These findings are broadly consistent with a previous qualitative study by Lee et al (9). Addressing the 3 Hypotheses Hypothesis 1 An increase in bitemporal VF loss with increasing chiasmal elevation was clearly demonstrated, but this result did not distinguish between the anatomical and crossing theories. Hypothesis 2 Although increasing chiasmal elevation resulted in increasing temporal VF loss, there was no increase in nasal VF loss. This finding is consistent with the crossing theory and not the anatomical theory. vulnerability of crossing fibers. However, a combination of any or all the above factors with the crossing theory (6) would explain the findings of the current study. There were several limitations of our findings. The first was that, despite identifying over 100 patients with pituitary compression, the number ultimately suitable for analysis was small and, in particular, the number of subjects with demonstrably eccentric compression was very small. Second, the coronal sections of the standard clinical MRI scans were relatively widely spaced, and this may have led to some inaccuracy in identifying the plane of the chiasm. However, this is unlikely to have affected the conclusions of the study. Third, the analysis of the pattern of chiasmal compression using only 3 points was very basic. More detailed analysis with finer MRI cuts and 3-dimensional surface analysis would, potentially, be more informative. Finally, the analysis of VF loss was based on an arbitrary allocation of numerical values to the statistical estimates of the pattern deviation plots. We are unaware of other studies that investigated the relationship between the site of chiasmal compression and resultant pattern of VF loss in detail. The findings presented here lend support to the crossing theory, but a larger, prospective study with greater variation in the eccentricity of chiasmal compression is warranted. Future investigations should refine image acquisition and, thereby, enable more precise definition of tumor eccentricity and chiasmal elevation. Optical coherence tomography of the retinal nerve fiber layer also might prove useful to supplement data derived from VF analysis. In addition to shedding light on the mechanism of neuronal damage resulting from compression, such studies would also inform future finite element models. STATEMENT OF AUTHORSHIP Category 1: a. conception and design: E. J. Kane, C. J. Lueck, D. E. Ashton, P. J. Mews, K. Reid, and A. Neely; b. acquisition of data: E. J. Kane, C. J. Lueck, D. E. Ashton, P. J. Mews, K. Reid, and A. Neely; c. analysis and interpretation of data: E. J. Kane, C. J. Lueck, D. E. Ashton, and A. Neely. Category 2: a. drafting the manuscript: E. J. Kane and C. J. Lueck; b. revising it for intellectual content: E. J. Kane, C. J. Lueck, D. E. Ashton, P. J. Mews, K. Reid, and A. Neely. Category 3: a. final approval of the completed manuscript: E. J. Kane, C. J. Lueck, D. E. Ashton, P. J. Mews, K. Reid, and A. Neely. Hypothesis 3 There was no indication of increase in ipsilateral nasal VF loss with increasing eccentricity. Again, this finding is consistent with the crossing theory but not the anatomical theory. Previous speculation (3-5) that nasal (crossing) fibers are selectively vulnerable simply and solely because of their anatomical location in the center of the chiasm is not supported by this study. Factors such as pressure (3), stretch (4), and ischemia (5) may all be involved, but on their own, they cannot explain the demonstrably increased Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 ACKNOWLEDGMENTS The authors thank MaryAnne Gregory, Senior Orthoptist, at the Canberra Hospital, for her assistance in retrieving the visual fields from the Humphrey Field Analyzer, and Dr. Jouke de Baar for advice on visual field and image analysis. 337 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution REFERENCES 1. Levin LA. Topical diagnosis of chiasmal and retrochiasmal disorders. In: Miller NR, Newman NJ, eds. Walsh and Hoyt's Clinical Neuro-Ophthalmology. 6th edition. Philadelphia, PA: Lippincott Williams & Wilkins, 2005:503-574; Chapter 12. 2. Kidd D. The optic chiasm. Clin Anat. 2014;27:1149-1158. 3. Kosmorsky GS, Dupps WJ Jr, Drake RL. Nonuniform pressure generation in the optic chiasm may explain bitemporal hemianopsia. Ophthalmology. 2008;115:560-565. 4. Hedges TR. Preservation of the upper nasal field in the chiasmal syndrome: an anatomic explanation. Trans Am Ophthalmol Soc. 1969;67:131-141. 5. Bergland R, Ray BS. The arterial supply of the human optic chiasm. J Neurosurg. 1969;31:327-334. 338 6. McIlwaine GG, Carrim ZI, Lueck CJ, Chrisp TM. A mechanical theory to account for bitemporal hemianopia from chiasmal compression. J Neuroophthalmol. 2005;25:40-43. 7. Wang X, Neely AJ, McIlwaine GG, Tahtali M, Lillicrap TP, Lueck CJ. Finite element modelling of optic chiasmal compression. J Neuroophthalmol. 2014;34:324-330. 8. Wang X, Neely AJ, McIlwaine GG, Lueck CJ. Multi-scale analysis of optic chiasmal compression by finite element modelling. J Biomech. 2014;47:2292-2299. 9. Lee IH, Miller NR, Zan E, Tavares F, Blitz AM, Sung H, Yousem DM, Boland MV. Visual defects in patients with pituitary adenomas: the myth of bitemporal hemianopsia. AJR Am J Roentgenol. 2015;205:W512-W518. 10. Larmonde A, Larmonde P. L'atteinte de genou posterieur du chiasma. Rev Otoneuroophtalmol. 1977;49:1-2. Kane et al: J Neuro-Ophthalmol 2019; 39: 333-338 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited.