Radiosurgery treatment of Anterior Visual Pathway Meningioma (AVPM)

Most anterior visual pathway meningiomas (AVPM) are benign and slow-growing, but these tumors may affect visual functions, including visual acuity (VA) and visual eld (VF). Due to location, most are treated non-surgically by fractionated stereotactic radiotherapy (FSRT), aiming to prevent tumor progression and visual functions deterioration. Unfortunately, FSRT in itself may affect visual functions. The current preferred treatment regimen (in terms of safety and effectiveness) is undetermined. While most cases are treated with conventional fractionation (cFSRT) – 50.4–54 Gy in 28–30 fractions of 1.8-2 Gy, advances in technology have allowed shortening of total treatment length to hypofractionation (hSRT) – 25-27Gy in 3–5 fractions of 5–9 Gy. Our aim was to evaluate the association of radiotherapy regimen for treating AVPM (cFSRT vs. hSRT) with visual function outcomes (VA, VF) at the last neuro-ophthalmologic evaluation. ONSM.


Introduction
Meningioma accounts for about one-third of adult brain tumors (population incidence 7.61 per 100,000).
The incidence increases with age and is higher in women. In most cases, this is a benign tumor (WHO -World Health Organization Grade I). Benign meningioma (as opposed to atypical (Grade II) or malignant (Grade III) meningioma) is not a signi cant cause of mortality but can cause severe morbidity 1 . Anterior visual pathway meningioma (AVPM) in most cases is benign and grows slowly over years but can impair one or more visual functions: visual acuity (VA), visual eld (VF), or color vision. The injury can be unilateral or bilateral depending on the tumor's location relative to the anterior visual pathway (AVP): near the optic tract, the optic chiasm, optic nerve, or involving the optic nerve sheath. The tumor may cause diplopia due to compression of cranial nerves III, IV and VI in the cavernous sinus or in the orbit 2,3 .
Without treatment, deterioration of vision functions 4 and nally blindness 5 occur.
In most cases, AVPM is diagnosed by imaging (without pathological veri cation) and is not operable. The common treatment is FSRT 6 , accurate radiation therapy designed to stop tumor growth and prevent deterioration of visual functions. The method's stereotactic component relates to using a patient-speci c immobilization system while assimilating a recent imaging test, thus producing a patient/tumor-speci c 3D coordinate system used throughout the treatment 7 . The optimal fractionation scheme -the daily dose of radiation and the total number of doses -for safe and effective treatment of AVPM is undetermined 8-12 . In our institution, most AVPM cases receive conventionally fractionated stereotactic radiotherapy (cFSRT), employing 28-30 daily doses of 1.8-2 Gy per day for a total dose of up to 54 Gy.
AVP structures exhibit greater sensitivity to single-fraction irradiation than other cranial nerves in the cavernous sinus, perhaps more so in patients with a pre-existing visual de cit due to tumors or previous surgery 13, 14 . In many AVPM cases, there is an impairment in visual functions (to varying degrees) even before treatment is given 4 . Radiation toxicity can damage optic pathways and impair their functions, particularly by radiation-induced optic neuropathy (RION). It is hypothesized that the mechanism of damage in RION is an ischemic disorder followed by necrosis of the optic nerve and the optic chiasm, typically occurring three months to several years after radiotherapy completion, with peak incidence after 1 to 1.5 years 15 . Important risk factors for RION include total radiation dose (over 50Gy), higher dose per radiation fraction, advanced age, previous exposure to chemotherapy, or optic nerve compromise at the beginning of radiation therapy. 15 No effective treatment for RION has been found, driving the need for careful consideration of radiation regimen, and raising the importance of estimating radiation toxicity risk for tumors adjacent to AVP. Other factors that have been described to in uence meningioma treatment outcome include residual tumor volume after surgery and pathology report of WHO grade, with atypical and malignant tumors entailing worse prognosis 16,17 .
Rogers et al. reviewed the treatment of meningioma and concluded that a total dose of up to 50Gy for optic nerve sheath meningioma (ONSM) and AVPM produced good results 16 . Stiebel-Kalish et al. reported their results and reviewed previous publications on cFSRT (1.7-2Gy / fraction, for a total of 50-60Gy) for AVPM 6 . In their study, tumor control was achieved in 14 of the 16 patients, with shrinkage of size in three patients during a mean follow-up of 39mo. Two cases progressed, one in an area that was outside the radiation eld. Visual function improved or stabilized in 8 of the 16 patients and worsened in 2 (12%). In published data, we found permanent deterioration in visual acuity or visual eld in only 23 out of 1191 patients after cFSRT (2.1%, see Additional File 1 for references). This rate corresponds to the risk of damage to optic nerves and chiasm summarized by Mayo et al. 18 , based mostly on descriptive publications with relatively small samples. The differences between published studies in outcome variables, therapeutic equipment, and follow-up periods make it di cult to reach conclusions, especially since some studies did not consider tumor progression (PD) as a variable that may affect the outcome.
Radiobiological models suggesting that higher daily doses are at least as effective as lower daily doses, and possibly more effective, while shortening treatment courses, led to abbreviated radiation regimens, termed hypofractionation (hSRT) 19 . Compared to cFSRT, hSRT uses higher daily radiation doses with shorter treatment duration. Choosing between these regimens for meningiomas in direct contact with the AVP is a double-edged sword: high-dose fractions reaching the sensitive blood supply of the optical system could lead to vascular damage and late secondary toxicity with loss of vision; on the other hand, a suboptimal dose could cause visual function deterioration as a result of PD.
Several series of patients with AVPM treated by hSRT have been published, with the caveat of insu cient reporting of dosimetric analysis for AVP structures or detailed measurement of visual function: Conti et al. 8 reviewed previous publications reporting hSRT treatment of AVPM and described "controlled tumor growth" and lack of "optic nerve toxicity" in a series of 25 patients treated with 2-5 doses of 4-10Gy each.
Hiniker et al. 20 summarized previous publications (including Emami et al. 21 and later information from QUANTEC 22 ) and reported results of treating peri-optic tumors by hSRT in up to ve fractions; they suggested both hSRT and SRS were safe treatment options. Marchetti et al. 23 reported the results of treating 143 patients with hSRT (25Gy in 5Gy sessions over ve consecutive days). VA change (worsening or improvement) was de ned as one or more Snellen lines. VF deterioration was de ned as an increase in the defect area. The authors reported a visual worsening rate of 7.4% (5.1% after excluding cases with PD). Conti et al. 12 reported a multicenter retrospective heterogeneous cohort of 341 patients with skull-base meningiomas, and compared cFSRT to hSRT. Visual toxicity was reported for one case (0.49%) of mild visual disturbance in hSRT vs. one case (0.7%) of moderate optical pathway toxicity in cFSRT. Most recently, Marchetti et al. 24 published results of 167 patients treated with 5X5Gy hSRT. The authors reported an overall visual worsening rate of 5.5%, or 3.7% if excluding PD patients with no details regarding the test used.
In light of the paucity of su ciently detailed information, the present study investigated the relationship between the radiation therapy regimen (hSRT vs. cFSRT) and the change in visual function (visual acuity and visual elds), measured at last neuro-ophthalmologic evaluation compared to pre-treatment.

Materials And Methods
We conducted a retrospective cohort study of patients with AVPM treated with hSRT or cFSRT at Sheba Medical Center during 2004-2015. After receiving the local IRB approval, the data were extracted from patients' computerized records. Statistical analysis was performed on anonymized data.

Patient population
Patients were entered into analysis according to the following inclusion criteria: 1) patient received a radiological diagnosis of meningioma. 2) the tumor involved or was in anatomical proximity to one or more of the following locations: medial sphenoid wing; cavernous sinus; orbital apex; optic nerve sheath; tuberculum sella. 3) the tumor was treated with radiotherapy using hSRT or cFSRT protocols at Sheba Medical Center during 2004-2015. 4) data on neuro-ophthalmology and neuroimaging were available. Exclusion criteria included: 1) Lack of visual acuity documentation in the involved eye before treatment, 2) patient underwent additional radiation therapy or surgery in the period between the rst radiotherapy treatment and the rst neuro-ophthalmologic assessment, 3) History of prior treatment with stereotactic radiosurgery (SRS), or 4) patient with no light perception (NLP) in the involved eye before treatment.
For each patient, we collected demographic variables (age at the time of treatment, gender, smoking history, and duration from symptom onset to diagnosis), pathology report of WHO grade of the tumor (if available), presenting signs and symptoms (headaches, seizures, blurred vision, ptosis, whether the tumor was an incidental nding) and medical history (previous surgery for the same tumor, prior radiation exposure, diagnosis of neuro bromatosis, vision-threatening systemic conditions such as diabetes, hypertension, collagen vascular disorders, chronic eye disease) ( Table 1).
We also reviewed data from the neuro-ophthalmologic examination before and after the radiotherapy.
Best corrected visual acuity was determined by a neuro-ophthalmologist using Snellen chart. Visual elds were examined by Humphrey automated static perimetry (HVF). The evaluation included optic disc abnormality (no/atrophy/swelling), other cranial nerve abnormalities (III, IV, V1, V2, V3, VI, VII), the diagnosis of RION, radiation retinopathy, and other radiotherapy complications.

Variable de nitions
Visual acuity (VA) is a continuous quantitative variable, determined according to the logarithm of the minimum angle of resolution (LogMAR) 25 . We used a value of 1/400 Snellen chart (LogMAR = 2.6) to represent counting ngers (CF) and estimated values of LogMAR 2.7, 2.8, 2.9 to represent hand movement (HM), light perception (LP), and no light perception (NLP), respectively 26 . We de ned variables related to change over the follow-up period to counteract the variance in pretreatment visual function evaluation. The VA change between pre-treatment evaluation and last evaluation was de ned as: ΔLogMAR = last LogMAR minus pre-treatment LogMAR. A difference of 0.2 LogMAR (two lines on a Snellen chart) was de ned as a clinically relevant change in VA.
The visual eld (VF) is a continuous quantitative variable determined according to the mean deviation (MD) value in the last assessment documented 25 . The VF change from the pre-treatment evaluation to the last evaluation was de ned as ΔMD = last MD minus pre-treatment MD. Radiotherapy treatment was de ned as hSRT if it involved ve fractions and as cFSRT if it involved 25-30 fractions. In cases of AVPM with bilateral effect, the worse eye (right or left) was determined according to the most recent neuro-ophthalmological assessment before treatment or by the side involved in the last pre-irradiation imaging examination. Visual function results were analyzed separately for "worse" and "better" eyes.
For patients who underwent further treatment after the cFSRT/hSRT (surgery or additional cranial reirradiation), visual acuity and visual elds were determined according to the most recent pre-treatment assessment.

Statistical analysis
Categorical variables were described as numbers and percentages. Continuous variables were reported as the median and interquartile (25%-75%) range (IQR). Categorical variables were compared between the groups using the chi-square test or Fisher's exact test as appropriate. Continuous variables were compared between the groups using the Mann-Whitney U test. Changes in continuous variables -between baseline and last assessment -were evaluated using Wilcoxon's signed-rank test. All statistical tests were two-sided. P-value < 0.05 was considered statistically signi cant. Statistical analysis was performed using SPSS software (IBM SPSS Statistics for Windows, version 25, IBM Corp., Armonk, NY, USA, 2017).

Radiation therapy data
Computerized records of radiotherapy treatments were retrieved, and data were collected regarding the date of treatment, the number of fractions, the daily radiation dose, and the total radiation dose.

Patient Inclusion
Of 288 patients with a radiological diagnosis of AVPM treated with radiotherapy at Sheba Medical Center

Patient Characteristics
There was no statistically signi cant difference between the groups in patients' demographic characteristics or medical background (Table 1). Eighteen patients showed evidence of optic nerve sheath involvement at baseline, 16 in cFSRT cohort and 2 in the hSRT cohort. Median follow up from radiotherapy to last neuro-ophthalmologic evaluation was 55 months (range 18-162 months for all patients), with a signi cantly longer follow-up in the cFSRT cohort: median 73mo (range 21-162 months) in the cFSRT cohort vs. median 37mo (range 18-75 months) in the hSRT cohort (p < 0.0001). A neuroophthalmologist reviewed the medical records of all the patients known to have a systemic visionthreatening condition (e.g., diabetes mellitus, hypertension): No patient had evidence of ocular damage secondary to a systemic condition. Pre-treatment data Neuro-ophthalmological evaluation prior to treatment revealed higher proportion of cFSRT patients with optic disc abnormality (atrophy/edema): 76% (25 of 33) compared with 27% (3 of 11) in the hSRT group (p = 0.009). The median pre-treatment better eye LogMAR was 0.10 in hSRT vs. 0.00 in cFSRT (p = 0.079).
There was no statistically signi cant difference between cFSRT and hSRT regarding pretreatment worse eye LogMAR (p = 0.925), worse eye MD (p = 0.530), or better eye MD (p = 0.173). Cranial Nerve (CN) V1 involvement was found in 33% (3 of 9) in the hSRT group compared to 4% (1 of 28) in the cFSRT group (p = 0.038). Three patients in the cFSRT group were diagnosed with chronic ocular disease, compared with no patients in hSRT, with no signi cant difference between the groups: see Table 2.
Analysis of MRI before treatment indicated that the proportion of patients with a proximity of the tumor to a midline structure was higher in the hSRT group, as was the proportion of ONSM, although these differences were not statistically signi cant (see Table 1).
Histopathologic classi cation (WHO tumor grade) was documented in 13 cases with WHO Grade I or II tumors with a signi cant difference between the groups (Table 1). 13 patients were classi ed as having "unknown histology" (radiologic diagnosis without pathologic veri cation): Nine patients (56%) in the cFSRT, compared with four patients (40%) in the hSRT group.

Radiotherapy schedule and dose
The number of radiotherapy fractions was ve in the hSRT cohort, compared with 28 in almost all patients in the cFSRT cohort. The daily dose in the hSRT cohort was 500 centiGray (cGy) compared with 180 cGy in the cFSRT cohort. Table 3 lists the radiation therapy characteristics.   (Table 2).

Change in visual acuity and visual eld between pretreatment and nal evaluation
The median change in visual acuity (ΔLogMAR) in the worse eye during the last neuro-ophthalmologic assessment compared to the pre-irradiation evaluation was 0.00 (IQR − 0.05, 0.10) in the cFSRT group and 0.05 (IQR 0.00, 0.50) in the hSRT group, suggesting a better long term outcome in the cFSRT cohort (p = 0.092). Of the 48 patients in the study, 12 (25%) had a clinically relevant deterioration in visual acuity (ΔLogMAR ≥ 0.2) in the involved eye in the last neuro-ophthalmologic assessment: six patients (17%) in the cFSRT cohort and six patients (46%) in the hSRT cohort (p = 0.061) ( Table 4).
In the hSRT cohort, the median LogMAR of worse eyes in the nal assessment was 0.30, compared with 0.10 at baseline, suggesting post-treatment VA deterioration after hSRT. However, this difference did not reach statistical signi cance (p = 0.068).
In the cFSRT cohort, the scores of the MD value of the worse eye in the last assessment were statistically signi cantly better than the scores of MD value of the worse eye at baseline (Wilcoxon Signed-Ranks test, p = 0.034). However, this result is based on only 16 cases (of 35), and it may be biased due to the remaining cases' missing data. cFSRT, conventionally fractionated stereotactic radiotherapy; hSRT, hypofractionated stereotactic radiotherapy; LogMAR, logarithm of minimum angle of resolution; MD, mean deviation * Wilcoxon's signed-rank test, two-tailed.The proportion of tumors involving the optic nerve sheath in the cFSRT cohort was 46%, compared with 15% in the hSRT cohort (p = 0.092). To control for this potential confounder, we re-analyzed the data for all variables in a subset of patients (hSRT, n = 11, cFSRT, n = 19) in which no optic nerve sheath involvement was documented in the pre-treatment imaging test.

Imaging ndings
Evidence of disease progression (PD) in imaging during the follow-up period was found in 5 patients (14%) in the cFSRT cohort, compared with 2 (15%) in the hSRT cohort (p > 0.999). Since the median duration from the last MRI scan to the last neuro-ophthalmologic assessment for all patients was two months, we believe that the imaging reports accurately re ect the tumor's state at the time of this evaluation.

Discussion
We studied patients treated with radiotherapy with a threat to visual function due to tumor location (in or near the AVP). The results are intended to support therapeutic decision-making in patients with AVPM not amenable to surgical treatment by pointing to a radiation therapy regimen associated with better posttreatment vision functions, to improve, even if only slightly, the prognosis of their vision. We examined the association between radiation regimen and visual function results and radiological PD, in patients with AVPM. Patients in the cFSRT group were more likely to have preserved VA after radiotherapy compared to hSRT group, although they had more optic disc abnormalities before and after the treatment ( Table 2).
The proportion of patients (25%) who had a clinically relevant VA deterioration was higher than we expected from previous publications on cFSRT (see Additional File 1) and hSRT 5,7,26−30 . This may be due to differences in measured and reported variables (mainly VA), follow-up length, prognostic factors such as histology, equipment, and radiotherapy treatment plans.
While there was no difference in pre-treatment visual function between patients assigned to each radiotherapy regimen, VA deterioration was more evident in hSRT than in cFSRT, although ndings did not reach statistical signi cance. This difference disappeared when we examined a subset of patients Of the 48 cases in the study, WHO grade was documented in 13 cases, mostly WHO grade I. We assume that the vast majority of other tumors was also WHO Grade I meningioma. In this cohort we nd that both radiation regimens were e cient in controlling AVPM in imaging, as 85% of patients were without PD at the last assessment, with no signi cant difference between hSRT and cFSRT at a median follow-up period of 4.5 years after treatment. This result agrees with previous publications stating that ve to ten years after radiation therapy, a local control rate of 68-100% was reached in WHO Grade I meningioma (presumed or histologically veri ed) 12,16,24 .
This study suffers several weaknesses due to its retrospective nature, including missing data for many patients, and the differences between study cohorts in treatment year and follow-up length. These may confound the results related to visual function and PD. However, unlike previous reports, our study has the strengths of using neuro-ophthalmologists' meticulous assessment of visual function and quantitative deterioration criterion, which allowed us to identify cases that may not have been identi ed if data relied solely on patient reporting. Other strengths include evaluating prognostic factors such as pathology reports that are not available in many studies, having long-term follow-up of the patients, and employing a pre-and post-treatment analytic design.

Conclusions
Although the study did not con rm a statistically signi cant association between the radiation therapy regimen (hSRT vs. cFSRT) and visual function outcomes in post-treatment years, our ndings using a comprehensive and meticulous investigation of visual outcomes suggest that cFSRT may be associated with less VA toxicity. Given the small sample and retrospective nature of our study, caution is needed in concluding which is the best regimen to be used. Based on our results, we chose to keep the traditional cFSRT regimen for ONSM and use hSRT only for tumors adjacent to the AVPM. We suggest this group of patients requires a multidisciplinary follow-up with meticulous pre-and post-treatment neuroophthalmologic evaluation, along with magnetic resonance imaging and patient-reported outcomes. Such information, along with well-designed prospective studies, may improve our understanding of the relationship between radiation regimen and long-term outcomes in AVPM. We believe our results indicate the importance of using thorough clinical investigation when adopting new radiation regimens. Consent for publication: Since the images in Fig. 2 are entirely unidenti able and there are no details on these individuals reported within the manuscript, we believe that consent for publication of images is not required.
Availability of data and materials: the datasets analyzed during the current study are available from the corresponding author on reasonable request.

Figure 1
Inclusion and exclusion criteria of the study population. 288 patients with a radiological diagnosis of AVPM were treated with radiotherapy at Sheba Medical Center between 2004-2015. We included patients with a radiological diagnosis of meningioma whose tumors were in de ned anatomical locations near the optic nerves and whose neuro-ophthalmology and neuroimaging data were available. We excluded patients having previous stereotactic radiosurgery treatment (SRS), lacking visual acuity documentation before treatment, having no light perception (NLP) prior to treatment, or undergoing additional treatment before the rst neuro-ophthalmologic assessment. hSRT, hypofractionated stereotactic radiotherapy; cFSRT, conventionally fractionated stereotactic radiotherapy.