The well-defined classic syndrome of chiasmal compression by an expanding sellar mass derives its anatomical basis from the differential retinotopic arrangement of fibers in the visual pathway. [10] The severity of VA and/or VF loss at presentation appears largely to depend on pituitary tumor size, chiasm morphology and the presence of signal changes within the optic nerve. [8, 11-14] The pathophysiology of visual loss in pituitary adenomas is complex, with insights derived from several clinical and laboratory studies. [15, 16] Mechanical stretch of the optic apparatus results in both electrophysiological disturbances in neuronal transmission as well as ischemic changes along the anterior visual pathways. In the early stages of optic nerve compression, focal demyelination results in conduction block. The optic nerve is more vulnerable to mechanical deformation as compared to the relatively larger and more robust chiasm. [5] Nevertheless, chiasm height strongly correlated with preoperative VF and TV scores in our study, supporting the findings of other authors. Kinking of the optic nerve along the bony margins of the optic canal also results in disturbances in the local microcirculation, and the intracanalicular segment of the optic nerve is particularly susceptible to hypoperfusion, resulting in a superimposed ischemic optic neuropathy. [3] In this study, ONKA and BA were significantly associated with visual function scores, in agreement with previous reports in the literature. [3, 5]
Significant improvements in visual function are reported in the vast majority of patients undergoing transsphenoidal surgery for pituitary adenomas. [17, 18] Nevertheless, a small proportion of patients fail to completely regain “normal” VA or VF, thereby emphasizing the importance of neuro-ophthalmological prognostication in perioperative patient counselling. Recovery of visual impairment after surgery is a dynamic and time-dependent process. Transsphenoidal decompression of the sella results in immediate to early reversal of the conduction block, and visual function thereafter continues to improve in a step-wise fashion through continued remyelination and restoration of axoplasmic flow. [16, 19, 20]
The primary outcome measure chosen for this study was visual outcome at 1 year, since clinically meaningful improvements in VA and VF usually stabilize within a few months after surgery. The postoperative recovery of VA does not necessary parallel that of VF21, and there it has been suggested that improvements in VF tend to be more delayed than the early gains seen in the domain of VA. [19, 22, 23] In a recent prospective study of a large institutional cohort, Wang et al. demonstrated no significant improvement in LogMAR scores of VA after surgery, while also detecting slower and delayed improvements in VF at 6 months postoperatively. [22]
The findings of our study suggest that re-alignment of optic nerve morphology and restoration of chiasm thickness at 3 months after transsphenoidal surgery is associated with VF scores at 1 year. In addition, patients with greater changes in BA and ONKA values benefited from greater enhancements in VF function. This is information allows for prognostication of long-term visual outcome. A number of studies have described a correlation between preoperative chiasm geometry and visual function, but few have analyzed dimensions of the chiasm after surgical decompression. Yoneoka et al. [4] showed that restoration of optic chiasm symmetry at 2 weeks and 3 months after surgery was associated with complete or near-complete recovery of VF deficits after surgery. More recently Zhang et al. reported no correlation between postoperative chiasm morphology and improvement in VF deficits. [24] However, in this study postoperative imaging was performed <72 hours after surgery which is probably an inadequate time-frame for any meaningful change in chiasm dimensions, especially in patients with chronic visual compression from a suprasellar mass. It may appear counter-intuitive that postoperative optic nerve deviations correlated with VF scores and not VA scores in our study. However, it is to be remembered that in this study VF scores were recorded for each individual eye. Given the frequently eccentric suprasellar growth patterns of pituitary adenomas, and variations in the position of the chiasm relative to the sella (typical, pre-, or post-fixed), varying degrees of concomitant optic nerve or optic tract involvement are expected, particularly in large tumors with pre- or retro-chiasmatic extension. This explains the improvement of VF with normalization of ONKA and BA values, reflecting relief of sagittal mechanical deformation of the pre-chiasmatic segment of the optic nerve.
Overall, VA scores did significantly improve after surgery in this cohort. In the early postoperative period, immediate reversal of the conduction block likely results in some improvements in VA. However, the lack of correlation between early postoperative improvements in OCD and 1-year VA scores, suggests an incongruity between anatomical and functional recovery, a phenomenon that has been noted by other authors. [26, 26] This discrepancy may, in part, derive from the multitude of factors governing reversal of visual dysfunction in these patients. From studies in patients with posterior ischemic optic neuropathy (PION), it is known that hypoperfusion may variably spare or involve the central part of the posterior optic nerve, through which runs the papillomacular bundle. [27, 28] The anatomical basis for these differences appears to relate to whether the blood supply to the nerve is derived from centripetal flow from the pial vascular plexus (making the central optic nerve a watershed zone) or, from axial centrifugal flow derived from intraneural branches of the central retinal artery. [28] Another factor confounding prognostication is the inter-relation between VA and VF. For example, a patient may learn to fixate eccentrically despite loss in the central 5 degrees of the VF, which may be interpreted as an improvement in VA. [28] We therefore used the TV in this study in an attempt to also provide a more integrated metric of visual function, however this approach may not accurately capture the extent of visual morbidity in these patients.
In this study, follow-up VA and TVS scores were poorer in patients who demonstrated chiasm sag on 1-year postoperative MRI. This finding, however, was probably related to differences in the baseline visual function scores and craniocaudal tumor dimensions. Chiasmal herniation after transsphenoidal resection has been demonstrated to cause late deteriorations in VF that may on occasion respond to surgical elevation of the chiasm (“chiasmapexy”). [29, 30] Although we report a fairly high incidence of radiologically evident chiasm sag in this cohort, none of our patients developed late onset deterioration of visual dysfunction. The clinical relevance of the chiasm sagging into the sella post-transsphenoidal surgery therefore remains questionable. [9]
Although beyond the scope of the current study, it is important to acknowledge that a number of other factors may offer useful prognostic information about visual outcome as well. Measurement of the retinal nerve fiber layer (RNFL) thickness using optic coherence tomography allows for an indirect estimate of axonal loss. [26, 31] Patients with thin preoperative RNFL tend to have suboptimal visual outcomes after surgery for pituitary tumors, and nasal RNFL thickness has been found to correlate with VF recovery. [32, 33] Intraoperative VEP monitoring as an adjunct to endoscopic transsphenoidal surgery may provide useful prognostic information with regard to VA and VF improvement. [34, 35] However, generation of a robust VEP signal is technically limited by its sensitivity to general anesthetic agents, and inconsistent use of threshold criteria. There is also evidence that assessing mean diffusivity and fractional anisotropy of the optic nerves using diffusion tensor imaging can predict visual outcome. [36, 37] However, lack of widespread availability and lengthy processing times limit the use of this modality in routine clinical practice.
Our study is limited by its retrospective nature, and by the fact that follow-up VA and VF data was not available in all patients. We are also unable to comment on long-term visual outcomes in our study cohort. Evaluation of optico-chiasmatic morphology was performed 3 months after surgery, since we hypothesized that re-alignment of the optic apparatus would stabilize by this time point. However, further evolution of these postoperative morphometric changes could continue to occur beyond this time frame. Further studies with more longitudinal imaging and clinical follow-up are needed to validate these findings.