Highlights
-
• Preoperative optical coherence tomography measurements may predict visual outcomes after pituitary neuroendocrine tumor resection.
-
• Preoperative retinal nerve fiber layer and ganglion cell complex thicknesses demonstrate similar strengths in predicting visual outcomes.
-
• Neuroradiologic variables could not reliably predict postoperative visual outcomes.
Introduction
Pituitary neuroendocrine tumors (PitNETs) are among the most common intracranial neoplasms, comprising a diverse group of tumors arising from the pituitary gland.Reference Scheithauer, Gaffey and Lloyd1 Although typically benign,Reference Ezzat, Asa and Couldwell2 they can cause significant morbidity through hormonal hypersecretion or local invasion.Reference Melmed, Kaiser and Lopes3 Vision loss is a common presenting feature and occurs through two main processes. Firstly, blockage of signal conduction at the optic chiasm can cause rapid but potentially reversible vision loss. Secondly, physical compression of the optic apparatus disrupts axo-plasmic flow via anterograde and retrograde degeneration of retinal ganglion cells with potentially irreversible blindness.Reference Moon and Shin4 Optical coherence tomography (OCT) is a non-invasive, non-ionizing imaging modality readily available in most ophthalmic practices that can visualize retrograde degeneration as thinning of the peripapillary retinal nerve fiber layer (RNFL) and macular ganglion cell complex (GCC).
The visual pathway can be decompressed through surgical resection of the tumor, but recovery of vision is variable and complicates the decision to undergo surgery in asymptomatic patients.Reference Gnanalingham, Bhattacharjee, Pennington, Ng and Mendoza5,Reference Muskens, Zamanipoor Najafabadi and Briceno6 Several factors have been proposed to influence visual outcomes, including age, severity of preoperative visual deficit, tumor size, RNFL thickness, and GCC thickness, with GCC thickness potentially being the most sensitive predictor.Reference Uy, Wilson, Kim, Prashant and Bergsneider7–Reference Tieger, Hedges and Ho12 GCC loss directly corresponds to axonal damage, and greater preoperative GCC thickness has been correlated with postoperative visual field (VF) improvement.Reference Tieger, Hedges and Ho12–Reference Yoo, Hwang, Yang, Joo, Kim and Kim14
MRI is a key tool for surgical planning. Preoperative tumor volume may predict final visual outcomes,Reference Uy, Wilson, Kim, Prashant and Bergsneider7 but other findings such as volume of the optic chiasm and extent of suprasellar invasion have been associated with visual recovery.Reference Hisanaga, Kakeda and Yamamoto15–Reference Luomaranta, Raappana, Saarela and Liinamaa17 To our knowledge, few studies have directly compared the relative prognostic value between OCT and MRI features of PitNETs to predict postoperative visual recovery. Several studies have demonstrated an association between thinner preoperative RNFL measurements and greater suprasellar extension and optic chiasm compression.Reference Glebauskiene, Liutkeviciene, Zlatkute, Kriauciuniene and Zaliuniene18,Reference Menon, Nair, Kodnani, Hegde, Nayak and Menon19 No studies have otherwise correlated MRI characteristics with ganglion cell analysis and investigated whether MRI or OCT findings more strongly predict visual outcomes.
Thus, the primary outcome of our study was to correlate MRI and OCT measurements with vision outcomes in patients with PitNETs who underwent transsphenoidal resection.
Methods
This retrospective cohort study was conducted by the Departments of Neurosurgery, Medical Imaging, and Ophthalmology & Vision Sciences with approval from the University of Toronto Research Ethics Board and adhered to the Declaration of Helsinki. Informed consent was waived given the retrospective nature of the study. Patients from a single tertiary neuro-ophthalmology clinic diagnosed with non-functioning PitNET through clinical presentation, hormonal profile and radiologic investigations were identified between January 1, 2008, and September 30, 2024. Inclusion criteria were non-secretory tumors approaching or contacting the optic chiasm as evidenced on dedicated MRI of the sella. Exclusion criteria included a history of previous pituitary surgery/radiotherapy or confounding ophthalmic disease affecting the optic nerve or retina apart from compressive optic neuropathy.
Demographic data collected included age and gender. Clinical measures included best-corrected visual acuity (VA), VF mean deviation (MD) measured by Humphrey 24-2 perimetry and the appearance of the optic nerve on fundoscopy (normal, temporal pallor, diffuse pallor and edema). Spectral-domain OCT was performed to measure the average thicknesses of the RNFL and of the nasal, temporal and overall GCC. All clinical measures were collected at initial presentation and 6 months postoperatively or sooner if no later follow-up was available following surgery. In those who did not undergo surgery, clinical measures were collected at initial presentation and at the 1-year follow-up visit to investigate outcomes longitudinally, and to avoid misclassifying patients who may have faced institutional circumstances that delayed their surgery (such as referral delay) into the non-surgical group. Positive VA outcome was defined as a final VA of 20/40 (logMAR 0.4) or greater as this threshold has been used in the literature to define none to minimal visual impairment.Reference Wu, Ji and Lin20–Reference Varma, Wu, Chong, Azen and Hays23 Visual acuities were converted from Snellen to logMAR for statistical analyses.Reference Tripathy and Gupta24 Positive VF outcome was defined as a final VF MD of −5 dB or better.Reference Zamanipoor Najafabadi, Genders and van Furth25 Measurements were taken from both eyes in all patients when appropriate.
MRI sella (voxel size: 0.5 × 0.5 × 2 mm) or brain MRI (voxel size: 1 × 1 × 1 mm) were acquired within 6 months preoperatively and at one year postoperatively or sooner. Using preoperative contrast images when available, otherwise T2-weighted images, a semi-automated method was used for tumor segmentation to obtain their preoperative volume (https://www.slicer.org/): First, the Otsu threshold was applied to isolate tumor regions by creating an optimal binary thresholded image that separates an image into foreground and background components,Reference Otsu26 then the grow from seeds module was used to generate a 3D segmentation, which was manually adjusted in all planes by author LP and then verified for accuracy by AB (board-certified neuroradiologist) (Figure 1). In addition to volume, maximum anteroposterior, craniocaudal and transverse dimensions of tumors were measured from coronal or sagittal sections. Postoperative tumor residual was defined as none (nonvisible), minimal (millimetric soft tissue) and gross residual (nodular obvious residual) on postoperative MRI scans. Semi-quantitative visual assessments of relationships with optic apparatus and the cavernous sinuses were performed for the pre-and postoperative MRI. For optic nerves (right and left) and chiasms, the mass was considered to be separate from, abutting or displacing them. Cavernous sinus invasion was defined as a Knosp classification of grade 3 or greater. Similar to tumor segmentations, these were generated by author LP and verified for accuracy by AB (board-certified neuroradiologist).
Representative MRI used to segment PitNET. (A) T2-weighted MRI sella with manually segmented adenoma (B, in red) showing coronal (main), sagittal (right) and axial (bottom) views. (C) Post-contrast whole brain MRI with manually segmented adenoma (D, in red), showing coronal (top left), sagittal (top right) and axial (bottom left) views.
Figure 1. Long description
A four-panel medical figure displays brain MRI scans in three planes: coronal, sagittal, and axial. Panels A and C show the original grayscale MRI images. Panels B and D repeat the corresponding images but add bright red overlays highlighting a central structure at the midline near the skull base, in the region of the hypothalamus/pituitary and upper brainstem. The highlighted area appears as a compact mass–like region across the different views. Panel C uses contrast-enhanced images, where several structures appear bright white. Orientation labels indicate R = right, S = superior, A = anterior, and P = posterior. The figure is intended to show the location of a particular anatomical region or abnormality across multiple MRI planes.
Descriptive and inferential statistics were conducted using IBM SPSS Statistics (ver. 29.0.1.0). Continuous variables were reported as mean values with a range/95% confidence intervals and/or standard deviation (SD). Analysis of variance was conducted to assess relationships between continuous data and VA/VF outcome. Chi-Square with Bonferroni correction was performed to identify associations between categorical variables. Pearson’s coefficient (r) and point biserial correlations were calculated to characterize correlations between study variables and VA/VF outcome, and Fisher’s z-transformation was used to identify significant differences in correlation coefficients. Receiver-Operating Characteristic (ROC) curve analysis was conducted to assess the utility of preoperative RNFL/GCC thickness in predicting VA and VF MD outcomes. The areas under the curve (AUCs) were calculated and compared using the methods described by Delong et al.Reference DeLong, DeLong and Clarke-Pearson27 Cutoff points maximizing the sensitivity and specificity were identified using the Youden Index (J).Reference Youden28 All figures were created in IBM SPSS Statistics and edited in GIMP.
Results
Sixty-eight patients were included in the study. Of these, 44 patients underwent transsphenoidal resection of PitNETs at the time of the study, and 24 were observed for at least one year. Of the entire cohort, 30 patients were female, and the mean age was 53.3 years (SD 1.89, range 20 to 85). Average follow-up interval was 97.0 weeks (SD 8.8, range 8.0 to 344.6). Further baseline clinical characteristics between groups (separated into surgical and non-surgical groups) are outlined in Table 1.
Clinical and MRI characteristics of patients identified with PitNETs approaching or contacting the chiasm. Mean values with 95% confidence intervals are presented for continuous variables unless otherwise stated. 36 preoperative MRIs were available to review
Table 1. Long description
This table compares clinical and MRI characteristics across three groups: All patients (n=68), Total underwent surgery (n=44), and No surgery (n=24). Section 1: Baseline Demographics & Vision Age: ˜53 years overall; median follow-up is 97 weeks. Ophthalmic Metrics: Evaluates Right, Left, and Worse eye for: Visual Acuity (logMAR) Visual Field Mean Deviation (VF MD) Retinal Nerve Fiber Layer (RNFL) thickness Overall, Nasal, and Temporal Ganglion Cell Complex (GCC) thickness Section 2: Initial MRI Features (n=36 overall; 34 surgery, 2 no surgery) Tumor Size: Median volume ˜11.5 cm3; max diameter ˜3.0 cm. Structural Involvement: Breaks down rates of approaching, abutment, and displacement/compression for the optic chiasm and optic nerve (right/left). Sinus Extension: Cavernous sinus extension (Yes/No) for right/left eyes. Postop Residual Tumor: None (16.7%), Trace (63.9%), Gross (19.4%).
RNFL = retinal nerve fiber layer; GCC = ganglion cell complex; VF MD = visual field mean deviation; VA = visual acuity; PitNETs = pituitary neuroendocrine tumors.
Of the 44 patients who underwent surgery, 36 had preoperative and 34 had postoperative MRI available to review (Table 2). Eyes achieving a positive VA outcome demonstrated better preoperative VA (0.3 vs 1.3, P = .02), VF MD (−8.1 vs −17.9 dB, P = .03), thicker RNFL (81.9 vs 69.5 µm, P = .01) and thicker overall GCC (69.0 vs 50.5 µm, P < .001), nasal GCC (64.6 vs 49.3 µm, P = .01) and temporal GCC (73.4 vs 52.5 µm, P < .001) sectors than eyes achieving a poor VA outcome. Eyes achieving a positive VF outcome also demonstrated better preoperative VA (0.3 vs 0.7, P = .05), VF MD (−7.4 vs −14.5 dB, P < .001), thicker RNFL (85.1 vs 70.5 µm, P < .001) and thicker overall (71.2 vs 55.6 µm, P < .001), nasal (66.5 vs 52.4 µm, P < .001) and temporal GCC (75.9 vs 59.4 µm, P < .001) sectors than eyes with a poor VF outcome. Smaller tumor volume (8.4 vs 18.2 cm3, P = .024) and younger age at surgery (49.1 years vs 60.3 years, P = .017) were also associated with positive VF outcomes. The rest of the MRI characteristics investigated, including optic nerve/chiasm involvement, residual tumor volume, maximal dimension of the tumor and invasion into the cavernous sinus, were not significantly associated with visual outcomes.
Clinical and MRI characteristics of patients identified with PitNETs approaching or contacting the chiasm. Mean values with 95% confidence intervals are presented for continuous variables unless otherwise stated. 36 preoperative MRIs were available to review
Table 2. Long description
This table presents comparative patient data split into two visual outcome analyses, including p-values for statistical significance. Top Section (Final Visual Acuity): Compares patients with “Final VA 0.3 or better” against “Final VA 0.4 or worse.” Patients with better final vision had significantly better preoperative vision (VA, VF, RNFL thickness, and GCC sectors, p>0.05). Bottom Section (Final Visual Field): Compares outcomes based on “Final VF −5dB or better” vs. “Final VF worse than −5dB.” Younger age, lower preoperative tumor volume, and better preoperative visual and structural metrics significantly correlated with better final VF outcomes (p>0.05). Anatomical & Imaging Features: Tracks tumor dimensions, optic nerve appearance (normal, temporal, or diffuse pallor), chiasm/optic nerve involvement, cavernous sinus extension, field defect types, and post-op residual tumor across cohorts.
*Significant at p < .05.
†Not significant after Bonferroni correction.
RNFL = retinal nerve fiber layer; GCA = ganglion cell layer thickness; GCC = ganglion cell complex; VF = visual field; VA = visual acuity; PitNETs = pituitary neuroendocrine tumors.
Both RNFL and GCC thickness correlated with positive VA (r = 0.28 (95% CI [0.068, 0.47]) P = .01; 0.41 (95% CI [0.18, 0.60]) P < .001) and positive VF outcomes (r = 0.44 (95% CI [0.24, 0.60]) P < .001; 0.57 (95% CI [0.37, 0.72]), P < .001) (Figure 2); however, equivalence testing did not reveal a significant difference between RNFL and GCC correlation coefficients (P = .4 and.3 for VA and VF respectively). Tumor volume significantly correlated with positive VF outcome (r = −0.39 (95% CI [−0.64, −0.056]), P = .024) but not with positive VA outcome (r = −0.33 (95% CI [−0.60, 0.0058]), P = .054) (Figure 3). Tumor volume also correlated with RNFL thickness (r= −0.45 (95% CI [−0.69, −0.11]), P = .011) and GCC thickness (r= −0.46 (95% CI [−0.73, −0.067]), P = .024).
Point biserial correlations between retinal nerve fiber layer (top) and ganglion cell complex thicknesses (bottom) with recovery of visual acuity to 20/40 and recovery of visual field mean deviation to −5 dB or better, in patients who underwent transsphenoidal resection of PitNETs. RNFL = retinal nerve fiber layer, GCC = ganglion cell complex.
Figure 2. Long description
The image shows four scatter plots arranged in a 2×2 grid, examining how presenting retinal nerve fiber layer (RNFL) and ganglion cell complex (GCC) thickness relates to final visual outcomes. Top left: Presenting RNFL thickness (µm) versus final visual acuity, divided into “20/50 or worse” and “20/40 or better.” Most points are in the better-vision group. The trend line slopes upward. Top right: Presenting RNFL thickness versus final visual-field mean deviation, comparing “worse than −5 dB” with “better than −5 dB.” Most points are again in the better-field group, with an upward trend. Bottom left: Presenting GCC thickness versus final visual acuity. Higher GCC values are predominantly seen in the better-vision group. Bottom right: Presenting GCC thickness versus final visual-field mean deviation. Higher GCC thickness is generally associated with better visual fields. Blue dots represent individual observations, and gray lines indicate the overall fitted trends.
Point biserial correlations between tumor volume with recovery of visual acuity to 20/40 and recovery of visual field mean deviation to −5 dB or better. Tumor volume was significantly associated with recovery of visual field only. RNFL = retinal nerve fiber layer; GCC = ganglion cell complex.
Figure 3. Long description
The image contains two vertically stacked scatter plots showing the relationship between preoperative tumor volume (cm3) and final visual outcomes. Top plot: The x-axis divides patients into “20/50 or worse” and “20/40 or better” final visual acuity. The y-axis shows tumor volume from 0 to 60 cm3. Blue dots represent individual patients. Most tumors are relatively small, but there are several larger tumors, including one near 57 cm3 in the better-vision group. A black trend line slopes downward from left to right. Bottom plot: The x-axis divides patients by final visual-field mean deviation: “Worse than −5 dB” versus “Better than −5 dB.” Tumor volume is again shown vertically. Most tumors are small, with a few larger outliers. The black trend line also slopes downward. Overall, both plots suggest a negative association between preoperative tumor volume and better final visual outcomes, although there is substantial variation among individual patients.
ROC curves for RNFL/GCC thickness predicting postoperative VA/VF outcomes are shown in Figure 4. All AUCs ranged between 0.8 and 0.86, except for the ROC curve of preoperative RNFL thickness predicting VA outcome with an AUC of 0.73. Differences between AUCs were not statistically significant (P’s > 0.1). Cutoff points of preoperative RNFL thickness were 73.5 µm (J max = 0.46) for predicting positive VA outcome and 77.0 µm (J max = 0.58) for predicting positive VF outcome. Cutoff points of preoperative GCC thickness were 56.5 µm (overall, J max = 0.71), 56.3 µm (nasal, J max = 0.47) and 64.5 µm (temporal, J max = 0.68) for predicting positive VA outcome and was 59.5 µm (overall, J max = 0.51), 51.5 µm (nasal, J max = 0.46) and 66.8 (temporal, J max = 0.51) for predicting positive VF MD outcome.
Receiver-operating characteristic (ROC) curves from preoperative RNFL/GCC and visual outcomes. (A) ROC curve predicting final VA 20/40 or better using preoperative RNFL thickness. Area under curve (AUC) = 0.73. (B) ROC curve predicting final VA 20/40 or better using preoperative GCC thickness. AUC = 0.86 (overall), 0.81 (nasal GCC), 0.86 (temporal GCC). (C) ROC curve predicting final VF −5 dB or better using preoperative RNFL thickness. AUC = 0.81. (D) ROC curve predicting final VF −5 dB or better using preoperative GCC thickness. AUC = 0.85 (overall), 0.81 (nasal GCC), 0.84 (temporal GCC). RNFL = retinal nerve fiber layer; GCC = ganglion cell complex; VA = visual acuity.
Figure 4. Long description
This image displays four Receiver Operating Characteristic (ROC) curve graphs labeled A through D in a 2×2 grid. Axes: Vertical axis is Sensitivity (0.0−1.0); horizontal axis is 1 − Specificity (0.0−1.0). Baseline: A gray diagonal line runs from (0,0) to (1,1) representing chance performance. Graph Layouts: Graph A (Top-Left): Single black line with vertical/horizontal steps, reaching ˜0.8 Sensitivity at 0.3 on the x-axis. Graph B (Bottom-Left): Three lines plotted: Overall GCC (black), Nasal GCC (blue), and Temporal GCC (green). All jump vertically at the origin to ˜0.5 Sensitivity before leveling off toward 1.0. Graph C (Top-Right): Single black line curving steeply upward, reaching 0.8 Sensitivity around 0.22 on the x-axis. Graph D (Bottom-Right): Three lines (Overall, Nasal, and Temporal GCC) starting around 0.4−0.6 Sensitivity at x=0, overlapping closely up to 1.0.
Among 22 patients with a preoperative VA of 20/50 or worse, preoperative RNFL thickness (r = 0.63 (95% CI [0.28, 0.83])) and preoperative GCC thickness (overall (r = 0.63 (95% CI [0.16, 0.87])), nasal (r = 0.55 (95% CI [0.023, 0.84])) and temporal (r = 0.69 (95% CI [0.25, 0.89]))) were correlated with an improvement in VA to 20/40 or better after surgery. Cutoff points predicting improvement to 20/40 or better were 60.0 µm (J max = 1.0) for preoperative RNFL thickness 53.0 µm (overall, J max = 0.92), 51.5 µm (nasal, J max = 0.83) and 54.8.5 µm (temporal, J max = 0.92) for preoperative GCC thickness. Among 35 patients with a preoperative VF of −5 dB or worse, preoperative RNFL thickness (r = 0.52 (95% CI [0.21, 0.73])) and preoperative GCC thickness (overall (r = 0.73 (95% CI [0.45, 0.88])), nasal (r = 0.64 (95% CI [0.30, 0.84])) and temporal (r = 0.74 (95% CI [0.47, 0.89]))) were correlated with an improvement in postoperative VF to −5 dB or better. RNFL and GCC (overall, nasal, temporal) thickness cutoff points predicting return of VF to −5 dB or better were 67.5 µm (J max = 0.57) and 61.0 µm (overall, J max = 1.0), 50.8 µm (nasal, J max = 0.86) and 69.8 µm (temporal, J max = 1.0), respectively. Multiple regression analyses did not identify a factor most predictive of VA or VF recovery in patients with an already poor preoperative VA or VF.
Of the 24 patients who did not have surgery, observation was recommended as the tumor was abutting the chiasm but not producing visual deficits in 12 and was approaching but not contacting the chiasm in 10. One patient requested observation, and one could not undergo surgery due to medical instability. In all patients who were observed, a joint decision was made between the patient, neurosurgical and neuro-ophthalmologic teams to maintain close follow-up with serial MRIs, VF/OCT imaging and visual assessments in accordance with local practice guidelines. At the most recent follow-up visit greater than one year after diagnosis, average VA in the worse eye was 0.11 (SD 0.11, range 0 to 0.3) and was significantly better than our predefined value of 0.4 for positive visual outcome (P <.001). Similarly, average change in VA in the worse eye was-0.01 (SD 0.13, range −0.40 to + 0.18), which was significantly better than what is considered a meaningful worsening in VA (> +0.2).Reference Rosser, Cousens, Murdoch, Fitzke and Laidlaw29 Average final VF MD in the worse eye in those who were observed was −3.24 dB (SD 2.90, range −12.88 to 0.57) which was significantly better than our predefined value of −5 dB for positive VF MD outcome (P = .005). Similarly, average change in VF MD in the worse eye was 1.16 dB (SD 3.6, range −3.03 to 11.38), which was significantly better than what is considered a meaningful worsening in VF MD (absolute decrease of >5 dB).Reference Patino, Varma, Azen, Conti, Nichol and McKean-Cowdin30 However, those who did not undergo surgery, in the worse seeing eye, had better baseline VA (0.13 vs 0.59, P = .003), VF MD (−4.26 vs −11.28 dB, P < .001), thicker RNFL (91.1 vs 76.9 µm, P = .002), thicker overall (78.2 vs 64.0 µm, P <.001), nasal GCC sectors (79.0 vs 60.3 µm, P < .001) and temporal GCC sectors (76.9 vs 67.4 µm, P = .02) compared to those who underwent surgery to begin with.
Discussion
This study investigated the clinical and MRI parameters associated with visual outcomes following resection of PitNETs approaching or contacting the chiasm. Similar to previous studies, preoperative VA, VF MD, RNFL/GCC thickness and tumor volume were associated with positive VA and VF outcomes. Even in those with poor preoperative vision, RNFL and GCC thickness are correlated with improvements in both VA and VF to a functional level. Neither measurement was more predictive of improvement in vision over the other. Neuroradiologic parameters were not predictive of visual outcomes.
Once chiasmal compression from PitNETs induces nerve atrophy, vision loss may be permanent. For this reason, RNFL thinning is strongly prognostic of poor visual recovery. Several studies have previously demonstrated the relationship between RNFL thickness and post-surgical visual recovery.Reference Danesh-Meyer, Wong and Papchenko31–Reference Han, Choi, Kim, Lee and Lee33 Ganglion cell inner plexiform layer thickness has also recently gained interest as a predictor of visual recovery, though results are mixed.Reference Blanch, Micieli, Oyesiku, Newman and Biousse11,Reference Han, Choi, Kim, Lee and Lee33–Reference Xia, Wenhui and Xiaowen35 Few, if any, large-scale studies have directly compared the prognostic value between RNFL and GCC thickness in patients with PitNETs.Reference Blanch, Micieli, Oyesiku, Newman and Biousse11,Reference Singha, Beniwal and Mailankody36,Reference Parija, Chakraborty, Sahu, Nayak and Rathod37 Our findings are in line with previous work which has demonstrated poor visual outcomes in patients with thinner preoperative RNFL and GCC thicknesses. Furthermore, our equivalence testing showed that the relative strength of the correlation between visual outcomes and RNFL versus GCC thicknesses is similar (Figures 2 and 4) and ROC analysis identified cutoff points of preoperative RNFL and GCC thicknesses in predicting VA/VF outcomes, suggesting that either may be used as a prognostic marker. RNFL and GCC thicknesses were also both correlated with an improvement in VA to 20/40 or better and/or an improvement of VF MD to −5 dB or better in patients with poor preoperative visual function (VA worse than 20/50 or VF worse than −5 dB) with neither one more predictive of visual recovery over the other. Though OCT imaging is commonplace in most ophthalmology clinics, ganglion cell analysis software may not be readily available, especially in low-income countries where the costs of an OCT machine with advanced software may be inaccessible. Therefore, basic RNFL analysis may provide similar prognostic capabilities to GCC analysis in predicting visual outcomes after surgical decompression of PitNETs.
RNFL thickness and tumor volume were correlated with one another as seen in previous studies.Reference Glebauskiene, Liutkeviciene, Zlatkute, Kriauciuniene and Zaliuniene18,Reference Menon, Nair, Kodnani, Hegde, Nayak and Menon19 Additionally, we show that GCC thickness is also correlated with tumor volume. Compression of the visual apparatus induces retrograde degeneration of retinal ganglion cells which becomes apparent as thinning of the macular GCC and peripapillary RNFL. While some preliminary studies have demonstrated higher sensitivity of GCC thinning in detecting chiasmal compression over RNFL thinning, future studies may further investigate whether chiasmal compression is detected earlier on GCC or RNFL analysis.Reference Agarwal, Jain and Singh38
Certain MRI findings such as pituitary tumor size, degree of suprasellar extension and volume of the optic chiasm have been reported as prognosticators of visual recovery.Reference Hisanaga, Kakeda and Yamamoto15–Reference Luomaranta, Raappana, Saarela and Liinamaa17 We demonstrated a correlation between tumor volume and positive VF outcomes, but not with positive VA outcomes. One explanation for this finding is that the shape of the tumor may affect VA more than its absolute volume. One study demonstrated that the tumor height above the sella may influence VA recovery more than overall tumor volume.Reference Uvelius, Valdemarsson, Bengzon, Hammar and Siesjö39 Furthermore, the soft consistency of PitNETs may not exert sufficient compressive forces on the macular fibers to cause dysfunction, depending on the tumor’s shape.Reference Li, Zhang and Ma40 We were also unable to find an association between perceived chiasm/optic nerve compression and visual outcomes. Therefore, sole visual interpretation of PitNETs, especially those which are atypically shaped, may be a limited predictor of visual outcome unless there is marked compression or apoplexy.
Surgery is often indicated when hormonal or visual deficits are present; however, asymptomatic large growing tumors may also be operated on to prevent future blindness.Reference Esposito, Olsson, Ragnarsson, Buchfelder, Skoglund and Johannsson41 Surgery itself can induce hormonal derangements and vision loss through damage to nearby structures; thus, the decision to operate must be balanced against the potential harms and benefits.Reference Molitch42 In our study, the decision to observe or pursue surgery was based on multiple factors, including but not limited to tumor size and its spatial relation with the optic chiasm, presence of visual deficits and optic atrophy, tumor growth rate and the medical status and wishes of the patient. Those who were not immediately scheduled for transsphenoidal resection were closely monitored by both neurosurgical and neuro-ophthalmologic teams, and the decision to undergo surgery was revisited at each follow-up. Of those who were observed throughout the study period for greater than one year, there was no significant worsening of visual function; however, these patients already demonstrated better VA and VFs at baseline, and one may argue that they were unlikely to develop significant decline in visual function given the slow-growing nature of PitNETs. It is also important to note that though our cohort did not suffer significant visual decline during the study period, it does not rule out vision loss at a later date past one year after diagnosis and thus these patients require long-term follow-up and evaluation. Further prospective studies with larger cohorts are required to determine whether deferring surgical resection of PitNETs can be advisable in certain cases.
There were several limitations to our study. Firstly, there were several factors that contributed to the relatively small sample size given the long study period. Our inclusion criteria for the study included only non-functioning macroadenomas which were approaching or abutting the chiasm without gross displacement of compression of the optic apparatus. Our goal was to identify patients with tumors that were already encroaching on or compressing the chiasm in order to determine the prognostic value of OCT/MRI parameters post-transsphenoidal decompression. Many of our MRI parameters were obtained using semi-quantitative measures to replicate standard-of-care neuroradiology interpretations. This may have limited our ability to draw statistical associations, for example, between visual outcomes and the relationship of the PitNET with the optic chiasm. Finally, we did not specifically investigate postoperative visual decline and the intraoperative factors that may influence vision loss after transsphenoidal resection as postoperative neuroimaging and clinical data were not available for all potential patients at the 1-year mark due to being lost to follow-up or being referred to another provider for ongoing care, lack of access to operative reports for all patients and insufficient patients who experienced visual decline greater than 0.2 logMAR or worsening of VF MD by at least 5 dB. Future multi-center studies with larger, more heterogenous cohorts, and longer follow-up periods may offer further information to guide clinicians and surgeons in their discussions with patients on balancing the risks of vision loss from surgery versus observation.
Conclusion
In summary, we demonstrated several predictors of VA and VF outcomes after transsphenoidal resection of PitNETs, including age, preoperative VA/VF MD, RNFL/GCC thickness and tumor volume. RNFL, GCC and tumor volume were similarly powerful prognostic factors for the prediction of visual outcomes, though tumor volume was significantly associated only with VF and not VA outcomes. The neuroradiologic relationship of the tumor with the optic apparatus and cavernous sinus was otherwise not associated with visual outcomes.
Author contributions
EM, AK and AB were responsible for conception of the project. MT, LP, YP, MM and RL were responsible for data curation. MT was responsible for data analysis and presentation. MT and LP were responsible for interpretation of the results. MT was responsible for initial drafting of the manuscript. All authors provided feedback to shape the results, figures and manuscripts.
Funding statement
No funding was acquired for the purposes of the study.
Competing interests
None of the authors have any relevant conflicts of interest to disclose.
Open access
