An Evidence-Based Vestibular Schwannoma Surgical Outcome Grading Scale for Large-to-Giant Vestibular Schwannomas


 Objectives: The objective of this study was to determine the relationship between facial nerve function and extent of resection (EOR) as outcomes in the surgical management of large vestibular schwannoma (VS) (≥ 2.5 cm maximal) and evaluate use of a new grading system that incorporates both outcomes.Methods: We conducted a systematic review of the electronic databases using different MeSH terms from 1990 to 2021. 5,623 patients from 56 studies were found appropriate for inclusion in the study. Surgical approach was reported in 5,144 patients, including translabyrinthine approach (TL) in 43.3% (n=2,225), retrosigmoid (RS) approach in 56.3% (n=2,899), retrolabyrinthine (RL) approach in 0.3% (n=16), and extended translabyrinthine (EX) approach in 0.1% (n=4). The proposed VS Grading System defines outcomes of gross total resection (GTR), near total resection (NTR) and subtotal resection (STR) with good facial nerve function [House-Brackmann (HB) Grade I-II] as Classes A, B and C respectively. Those with poor facial nerve outcome (HB III-VI) are graded as Classes D, E and F, respectively. Results:. As expected, patients with STR had a higher likelihood of better facial nerve outcomes (HB I-II) compared to NTR [Odds Ratio (OR): 7.30, 95%CI: 2.45-12.1, p=0.004] and GTR (OR: 9.61, 95%CI: 3.61-15.6, p=0.002), while NTR had better facial nerve outcome than GTR (OR: 1.5, 95%CI: 1.14-4.6, p<0.0001). A Class A result, representing the best possible outcome, was obtained in 55.8% of TL vs. 49.4% undergoing RS approach. Conclusion: Complete surgical resection with preservation of facial nerve function is the gold standard for large VS.


Introduction
Recently, the paradigm for vestibular schwannoma (VS) surgical treatment has been shifting towards what is sometimes referred to as a "nerve-centered" approach, in which preservation of facial nerve function (and, if possible, hearing) is the most important determinant of good versus poor outcome after surgery. [1][2][3][4][5][6][7] With the advancement in microsurgical techniques and neurophysiological monitoring, surgical mortality has been reduced to as low as 0.4% ,with anatomical preservation of facial nerve approaching 100%. [8][9][10] Although gross total resection (GTR) is the treatment of choice, near total resection (NTR) or subtotal resection (STR) is warranted in situations where facial nerve function is jeopardized by aggressive resection. [11] Therefore, there is a need to weigh the risk of future tumor growth against the bene t of facial nerve function in terms of extent of resection (EOR). The ideal EOR was historically controversial as Walter Dandy [12] opted for total resection of VS while Harvey Cushing [13] advocated partial resection to reduce morbidity. Further, William House reported in 1968 [14] that partial tumor removal was a reasonable alternative to total removal among patients who are elderly, have unstable vital signs, or are a poor surgical risk.
We aimed to address how to best evaluate surgical outcomes following resection of large to giant VS in particular. We believe, as do most other authors, that minimizing the risk of injury to facial nerve function is the paramount concern for the VS surgeon, and a post-operative facial nerve palsy is, generally speaking, incompatible with a good outcome overall. It follows that factors that expose the nerve to elevated risk should be avoided when possible. Studies have shown that EOR (usually categorized as GTR, NTR, and STR) is an independent risk factor for facial nerve injury, with the lowest risks of injury associated with STR in most studies [15][16][17].
What concerns us here is that EOR is most often treated as an independent risk factor for facial nerve outcome but not as a clinically relevant outcome in and of itself. In addition to the risk of facial nerve injury at the time of initial treatment, there is a risk of tumor recurrence with less than total removal, and, if severe enough, requirement for re-treatment with the consequent additional risk of facial or other nerve injury. Hence, the ideal surgical outcome is de ned as that in which both risk of injury to facial nerve and tumor recurrence are minimized. Previous studies note that while STR is associated with good facial nerve outcome, it is also associated with higher tumor recurrence rates. [11], [18] Other studies have also shown that EOR is an independent risk factor for tumor recurrence. [11,[15][16][17] We contend that because EOR is an important proxy for the risk of recurrence, it should be considered as a surgical outcome, albeit secondary in importance to risk of facial nerve injury at the time of treatment.
We performed a systematic review and meta-analysis incorporating large, pooled data sets using double evaluation methods (coupling facial nerve outcome and EOR). The main determinant for good outcomes remains facial nerve function, but our proposed outcome scale distinguishes good surgical outcomes (good facial nerve function and STR) from better and best outcomes (good facial nerve function and NTR, vs GTR, respectively), as well as bad outcomes (GTR with poor facial nerve function) from "worse" and "worst" (NTR and STR with poor facial nerve function, respectively). Based on available data, this meta-analysis demonstrates that a surgical outcome grading scale using both facial nerve function and EOR as one integrated outcome may be useful for setting standard outcome reporting for patients undergoing VS surgery.

Data search strategy:
We followed the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) guidelines for literature search. [19] A detailed systematic review was performed of electronic databases using PubMed, Google Scholar, EMBASE, Medline, The Cochrane library, Scopus, CINAHL, Primary studies and FDA reports, bibliographies of the key articles, and ClinicalTrials.gov for articles published in the English language between January 1st, 1990 and August 1st, 2021.MeSH terms (using the Boolean operators "and" and "or") were used.

Inclusion and Exclusion Criteria:
The inclusion and exclusion criterion were based upon the PICOS framework which included, (i) Types of participants: Adult patients who underwent VS surgery for large to giant tumor≥2.5cm, (ii) Interventions: Surgical treatment of VS using a variety of surgical approaches, (iii) Control: Not Applicable, (iv) Outcome measures: (a) Facial nerve function graded by the House-Brackmann (HB) scale at 1-year follow-up, (b) EOR (GTR, NTR and STR). Randomized controlled trials (RCTs), prospective cohort studies (PCS), retrospective comparative studies (RCS) and retrospective series were included in our analysis. Case reports and studies without a facial nerve outcome according to extent of resection were excluded. The study components along with inclusion and exclusion criterion are detailed in Table 1.

Data Extraction and Outcome measures:
The data were extracted using a structured template form based on the Cochrane Consumers and Communication Group. The meta-analysis was conducted in accordance with the criteria set by the Cochrane Consumers and Communication Group reviews: Meta-analysis. Both authors reviewed each article, and any disagreement among the authors was resolved by discussion. The following data were extracted from each article: (i) demographic characteristics, (ii) type of surgery (translabyrinthine, retrosigmoid, and others), (iii) extent of resection, and (iv) HB grade facial nerve outcome..

Statistical analysis:
Logistic regression analysis was used to measure the odds ratio (OR) and 95% con dence interval (CI) using random effect model. All tests were 2-tailed, and p value <0.05 was considered statistically signi cant. SPSS 23 was used to conduct independent sample t-tests and Spearman's coe cient correlation. For the establishment and validation of new grading system, z-test for the difference between two proportions was carried out.

Risk of bias across studies:
The Cochrane Collaboration's tool was used to assess the risk of bias across the included studies, including symmetry of the funnel plots and the Egger test to evaluate publication bias. A non-parametric trim-and-ll procedure identi ed and corrected for funnel plot asymmetry and re-estimated the aggregate results. The high heterogeneity among studies was further analyzed using funnel plots, which showed asymmetrical distribution. This bias can be attributable to sample size as the removal of smaller-sized cohorts signi cantly decreased heterogeneity.

Study Selection:
A total of 335 articles were retrieved from the various electronic databases (see Search Strategy in Methods) and reviewed according to PRISMA guidelines. After screening the abstracts, 256 articles were excluded as the data were not related to surgical treatment of large VS or did not present quantitative results. A total of 79 full text articles were assessed for eligibility, of which 23 were excluded based upon the inclusion and exclusion criteria. The remaining 56 articles were included in this review, including non-RCTs [20] , [8, 15-17, 21-52, 52-71].
Use of the Grading System to compare surgical approaches: We used the Vestibular Schwannoma Surgical Outcome Grading Scale to perform a non-weighted pooling of cases across those 40 of the 56 studies that included only one surgical approach or the other and used a simple statistical test of the difference between proportions. Overall, TL resulted in good HB function in 64.4% and RS in 57.1% (Z=4.12, p ≤ .0001). GTR was performed slightly more often by surgeons using RS than TL (91.9% vs 89.6%; Z=2.23, p ≤ .026). Table 3 presents the results using the grading system that combines EOR and HB facial nerve grade. For all six Grade outcomes, there was a signi cant difference between TL and RS. TL cases showed a higher rate of Grade A and Grade B results, the two best outcomes, while RS had a higher rate of Grade C outcomes as well as Grade D outcomes.
Although percent of cases with Grades E or F were quite small, RS showed slightly lower rates for these two poorest results categories. That is, when a combination of extent of resection and facial nerve function is considered, TL outperformed RS in providing a good facial nerve function outcome with the greatest extent of resection. A further look showed that when GTR was performed, TL had a rate of good HB of 62.2% compared to RS with a rate of 53.7% (Z=4.52, p ≤ .0001). Although GTR was reported more often when using RS than TL, good facial nerve results with GTR were more likely with the TL surgical approach. When STR was performed, RS outperformed for good HB result (94.5% vs 74.5%, respectively; Z=4.11, p ≤ .0001). The difference for NTR was not signi cant.  We found some evidence of publication bias, as suggested by slight asymmetry of the funnel plot (Egger test, z=-2.48; p = 0.009) and association between effect sizes and corresponding sampling variances (Begg test, z=-1.49; p = 0.04). According to the trim-and-ll method to correct for publication bias, the association in terms of clinical outcome was still signi cant after imputing 2 possible missing studies (adjusted OR: 6.18; 95% CI, 3.46-11.9; p <0.0001), suggesting no potential role for small-study effects or publication bias on the meta-analysis results.

Discussion
The ideal treatment for large VS is complete surgical removal with preservation of facial nerve function. However, these large tumors pose challenges and increase postoperative complication rates compared to smaller VS because patients may present with intracranial hypertension, multiple cranial nerve de cits, and/or signs of brainstem/cerebellum compression. Therefore, the choice of EOR for large VS remains unsettled. Most published studies use facial nerve function alone as the measure of outcome. [33,41,45,53] However, the current study presents and shows the usefulness of a novel surgical outcome scoring system incorporating both EOR and facial nerve function for evaluating the functional outcome in VS surgery.
STR is usually adopted as a nerve-centered approach, hence mostly performed in patients having large tumors adherent to the facial nerve. [72] However, patients with STR have a higher likelihood of tumor regrowth compared to NTR or GTR. [73,74] GTR is considered to be optimal to prevent tumor recurrence, while NTR also decreases recurrence rate due to the fact that it reduces blood supply or removes critical tumor mass to prevent regrowth. Therefore, the Vestibular Schwannoma Surgical Outcome Grading Scale presented here takes this into account in de ning outcome and may provide a standard method of evaluating surgical outcome among patients with VS worldwide.  [42] reported excellent facial nerve outcome in 75% of patients following GTR of large VS. Our results are consistent with these and other prior reports in that STR is associated with better facial nerve outcomes when looked at in isolation.
If we consider, EOR and HB facial nerve grade as independent variables, patients undergoing STR had a higher rate of good facial nerve outcome (61.0%) (Grade C) than NTR (59.9%) (Grade B) and GTR (52.2%) (Grade A). This concurs with a previously conducted meta-analysis by Gurgel et al [11] which found that STR was better than NTR and NTR was signi cantly better than GTR with regard to facial nerve outcomes. These results, as well as those in the current study, may be confounded by pre-operative HB grade, tumor size, cranial nerve de cits, adherence to facial nerve and other factors for which data were not available by individual patient for the meta-analysis. However, use of a grading system that combines HB grade with EOR can provide a different perspective on overall outcome.
There also remains a question about whether choice of surgical approach makes a difference for optimal facial nerve outcome [11]. The current meta-analysis suggested that the two most common approaches, translabyrinthine and retrosigmoid, were not signi cantly different when assessing facial nerve outcome or EOR alone. However, a difference could be shown using the new grading system. A further nonweighted pooling of cases across those 40 studies that included only one surgical approach or the other and use of a simple statistical test of the difference between proportions also showed the usefulness of the grading system, with TL outperforming RS in obtaining the two best outcome grades. Results might also be interpreted to suggest that if the surgeon is planning to do a subtotal resection, which is inherently less likely to damage the facial nerve, RS should have a higher proportion of good facial nerve outcome, but if planning to attempt to remove the entire tumor TL will yield better facial nerve results.
That is, surgeon intention regarding EOR is a confounder for facial nerve outcome as those not intending on trying to remove the whole tumor will generally be less inclined to go through the longer and more demanding translab approach. A signi cant caveat here is surgeon/inherent bias as these 40 studies included only surgeries performed by surgeons who preferred and used one surgical technique over the other, so that the differences between surgical approaches is confounded by possible differences between surgeons, their experience, and even possibly their patient populations.
A meta-analysis pools data from different studies and because of that there are a number of potential limitations: (i) inherent bias of the surgeons regarding EOR as well as surgical approach, (ii) variability in tumor size, (iii) degree of resection, (iv) length of follow-up, (v) varying surgical protocols among surgeons worldwide, and (vi) heterogenous populations. In addition, as noted above, it may not be possible in a meta-analysis to account for confounding factors such as pre-operative facial nerve grade, cranial nerve de cits, adherence of tumor to facial nerve or other individual patient risk factors. However, these are issues that will likely always remain problematic in trying to compare results across different studies of very large VS. Our proposed Acoustic Tumor Grading System would lessen some of the variability in reporting results while taking into account the two most important measures of success.

Conclusion
Complete surgical resection with preservation of facial nerve is the gold standard for large VS. However, a nerve-centered approach using near total and subtotal extent of resection is gaining in prominence. This new proposed grading system may help reduce inconsistencies in reporting of outcomes and help when strategizing management of such patients. Electronic Database Search Strategy using PRISMA Guideline