The effect of metformin on vestibular schwannoma growth: a systematic review and meta-analysis

Abstract

Objectives

To systematically review and evaluate metformin’s potential impact on vestibular schwannoma (VS) growth.

Data sources

PubMed, Cochrane Library, and Embase

Review Methods

Retrospective cohort study was performed on sporadic VS patients undergoing initial observation who had at least two magnetic resonance imaging (MRI) studies. Patients were stratified by metformin use during the observation period. Primary endpoint was VS growth, defined as at least 2mm increase in diameter. Survival free of tumor growth was evaluated between groups. Systematic review and meta-analysis were performed to produce a pooled odds ratio [OR]. Study heterogeneity was assessed and post-hoc power analysis performed.

Results

Total of 123 patients were included, of which 17% were taking metformin. Median patient age was 56.6 years (range, 25.1-84.5). There were no statistically significant differences between groups. Survival analysis did not demonstrate a statistically significant difference in time to VS growth between groups (hazard ratio=0.61, 95% confidence interval [CI]=0.29-1.29). Furthermore, logistic regression analysis did not demonstrate a statistically significant difference between groups in the odds of VS growth (OR=0.46, 95% CI=0.17-1.27). Systematic review identified 3 studies. Meta-analysis suggested that metformin reduces the odds of developing VS growth (pooled OR=0.45, 95% CI=0.29-0.71). Studies demonstrated low between-study heterogeneity. Power analysis demonstrated a sample size of 220 patients with equal randomization would be required to prospectively identify a true difference with 80% power.

Conclusions

Metformin use may reduce the odds of VS growth. A randomized trial would be ideal to identify an unbiased estimate of metformin’s effect on VS growth.

Lay summary

This article looks at whether taking metformin has an effect on slowing vestibular schwannoma growth by evaluating our own data and reviewing prior articles. In summary, taking metformin may decrease the odds of vestibular schwannoma growth.

Introduction

Vestibular schwannomas (VS) are benign tumors associated with unregulated Schwann cell growth surrounding the vestibulocochlear nerve. They are the most common tumor of the cerebellopontine angle and have been found to have an increasing incidence rate primarily conferred by increased access and utilization of magnetic resonance imaging (MRI)^1,2. As an increasing proportion of tumors are incidentally detected at a small size, upfront active observation for initial management has concomitantly increased, though decision making takes multiple factors into account^3,4. As such, identification of factors which either predict or impede VS growth during this observation period has become a significant interest.

Numerous studies attempting to qualitatively and quantitatively evaluate risk factors for VS growth on a population basis have provided some insight as to which may play a role, namely initial tumor size, non-incidental diagnosis, and disequilibrium at presentation^5–7. However, identification of factors which may prevent or impede VS growth remains even more elusive. Over the last two decades, multiple pharmacotherapies have been trialed, mostly in the neurofibromatosis type 2 (NF2) population, but none have demonstrated clear benefit^8,9. Bevacizumab and aspirin have been amongst the most promising, but ultimately have failed to demonstrate consistent effects in more thorough reviews^10–13. Similarly, studies evaluating the effects of metformin on sporadic VS growth have been variable^14–16. As such, the objective of this study was to contribute another single institution’s experience of the effects of metformin on VS growth, and then to systematically review and meta-analyze the literature to perform a pooled analysis to provide additional insight into metformin’s true effect on sporadic VS growth.

Methods

Retrospective cohort data acquisition

The University of Texas MD Anderson Cancer Center Institutional Review Board approval was obtained prior to initiation of the study (#PA19-0106). All patients were identified through an ongoing maintained database of VS patients between February 2000 to May 2022. Subjects were included if they presented with radiographic diagnosis of a suspected sporadic VS, underwent initial observation, and had at least two serial brain MRI scans with a minimum of 3 months apart. Exclusion criteria included those who underwent prior microsurgery or radiosurgery, had less than two MRI scans, and those VS associated with suspected NF1 or NF2. Subjects were then stratified by use of metformin during the observation period, which was defined as patients documented of taking metformin at the time of initial clinic presentation. The initial daily dose of metformin was recorded for these individuals.

Patient and tumor data were retrospectively collected, which included age at initial presentation, sex, race, initial tumor size, and length of observation. Tumor size was evaluated by width and length, which were defined as the largest tumor diameter using two-dimensional measurements in the anteroposterior and mediolateral planes, respectively, incorporating both intracanalicular and extracanalicular tumor components. The primary endpoint was VS growth, which was defined as at least 2mm increase in either measurement between the first and last MRI studies, in accordance with previous studies and what is readily used in generalized clinical practice^17,18. Length of observation was defined as time between the date of initial MRI and date of first MRI denoting tumor growth or date of most recent MRI. If the patient underwent subsequent microsurgery or radiosurgery, the most recent MRI scan was defined as the last MRI immediately preceding intervention.

Systematic review

Systematic review of the current literature was performed in order to identify all reviews evaluating metformin and VS growth and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines¹⁹. To identify studies for inclusion, a keyword search of PubMed, Cochrane Library, and Embase was performed using the following: “(‘metformin’/exp OR metformin) AND (‘vestibular schwannoma’/exp OR ‘vestibular schwannoma’ OR (vestibular AND (‘schwannoma’/exp OR schwannoma)))” and “(‘metformin’/exp OR metformin) AND (‘acoustic neuroma’/exp OR ‘acoustic neuroma’ OR (acoustic AND (‘neuroma’/exp OR neuroma))).” The databases were searched from inception through September 25, 2022. To identify additional articles for inclusion, the reference lists of relevant articles were hand-searched. References were exported to EndNote (Clarivate Analytics, Philadelphia, PA, USA) and screened for relevance.

After duplicates were removed, 8 articles were identified. Titles and abstracts were first used to screen for relevance. Five full-text articles were then reviewed for eligibility. Inclusion criteria included a retrospective or prospective cohort study design which evaluated VS tumor growth in participants with sporadic VS who underwent initial observation and had at least 2 MRI scans. Intervention inclusion criteria was the use of metformin during the observation period. These participants were compared to similar participants from the same studies who did not use metformin. The outcome of interest was the impact of metformin use on tumor growth. Exclusion criteria included non-cohort study, abstract only, and included patients who underwent prior microsurgery or radiosurgery, had less than two MRI scans, or had VS associated with suspected NF1 or NF2. All manuscripts were assessed by 2 authors (B.D.L. and P.W.G.) to identify all studies meeting inclusion and exclusion criteria. Any discrepancy that arose was solved with discussion until consensus was obtained. Data regarding number of patients with and without metformin use, demographics, follow up, tumor characteristics, approach to tumor measurement, definition of tumor growth, statistical analysis, and tumor growth outcomes were obtained. Articles were then critically appraised to assess the level of evidence using the Oxford Center for Evidence-Based Medicine criteria²⁰. The risk of bias was also critically assessed using the Joanna Briggs Institute Critical Appraisal Checklist for Cohort Studies²¹. An appraisal was performed for each study according to the checklist and an overall appraisal was determined. Eligible publications were reviewed to extract information regarding patient information, measurement of growth, statistical analysis, and outcomes. This search strategy is outlined in Figure 1.

Figure 1.

Detailed preferred reporting items for systematic reviews and meta-analyses flow-diagram.

Statistical analysis: retrospective cohort

Results were expressed as means for continuous variables and percentages for categorical variables. Fisher’s exact was used to compare groups for all categorical variables, while the Wilcoxon rank-sum test was used to compare groups for all continuous variables. A multivariable logistic regression model was fitted to assess the association between metformin use and tumor growth while including other covariates of interest. Backward selection was used to identify a final model, which was consistent with metformin use as the only covariate. Survival free of tumor growth was then evaluated using the Kaplan-Meier method, with time to event defined as time from date of initial MRI to date of first MRI denoting tumor growth or death, whichever occurred first. Patients alive without tumor growth were censored at last follow-up MRI. The Cox proportional hazards regression was used to compare metformin and non-metformin groups. P values <0.05 were considered statistically significant.

Statistical analysis: systematic review

Systematic review of the literature identified 3 studies evaluating the effect of metformin on VS growth. Only one study¹⁶ reported a hazard ratio (HR) from a time-to-growth model while 2 studies^14,15 reported odds ratios (OR) from evaluating VS growth as a binary outcome. A pooled OR was then obtained by meta-analyzing the results from two previous studies^14,15 in conjunction with the results from the study herein with a fixed-effects model due to the lack of heterogeneity. The meta-analysis summarized data using the OR and its corresponding 95% confidence interval (CI).

Heterogeneity across studies was evaluated using Cochran’s Q-test. Higgins I² statistic was then used to calculate the percentage of variation in the effect sizes across studies due to heterogeneity, where 0% indicates no observed heterogeneity and larger values indicate increasing heterogeneity. Finally, using the current study data, a two-sided Chi-squared test of equal proportions was used to calculate a post-hoc power analysis based on the same growth criteria used in this retrospective cohort study to direct future studies. All analyses were conducted using R version 4.1.2 and Stata v16 (StataCorp LLC, College Station, TX).

Results

Retrospective cohort

A total of 123 patients were included, of which 17% (n=21) were taking metformin. The mean patient age was 57.9 years and mean length of observation was 3.55 years. Slightly more patients were female (55.3%), and most patients were white (82.1%). There was no statistically significant difference between metformin and non-metformin groups regarding patient age, gender, race, length of observation, and initial tumor measurements. Demographic data and tumor characteristics are presented in Table 1.

Table 1.

Demographic and tumor characteristics between metformin and non-metformin users in the retrospective cohort study.

	No metformin	Metformin	Total	p value
Total patients	102	21	123	NA
Mean age (years)	57.2	60.9	57.9	0.256
Male (%)	43 (42)	12 (57)	55 (45)	0.235
Caucasian (%)	85 (83)	16 (76)	101 (82)	0.378
Mean length of observation (years)	3.47	3.92	3.55	0.861
Mean initial length (mm)	13.0	11.6	12.8	0.396
Mean initial width (mm)	8.5	8.0	8.4	0.749
Tumor growth (%)	53 (52)	7 (33)	60 (49)	0.130

NA: not applicable, mm: millimeters

Overall, 60 patients (49%) experienced tumor growth throughout follow up. Survival free of tumor growth at 1, 3, 5, and 10 years was 87.3, 59.8, 39.0, and 30.2%. Survival analysis did not demonstrate a statistically significant difference in time to VS growth between metformin and non-metformin groups (HR=0.61, 95% CI=0.29-1.29, p=0.19). Figure 2 demonstrates a Kaplan-Meier plot of survival free of tumor growth for both groups.

Figure 2.

Kaplan-Meier curve of survival free of vestibular schwannoma growth for metformin and non-metformin groups. Number at risk denotes the remaining patients in each group who had not demonstrated tumor growth by the according timepoint.

As there were no statistically significant covariates to include in the multivariable model, the final logistic regression model consisted of metformin group as the only covariate. Logistic regression analysis with this single covariate did not demonstrate a statistically significant difference between groups with regards to risk of tumor growth (OR=0.46, 95% CI=0.17-1.27, p=0.13). Moreover, a dose-effect was not observed as there was no significant association between metformin daily dosage and tumor growth (p=0.307).

Systematic review and meta-analysis

Study characteristics and data regarding demographics and tumor characteristics for the studies in the systematic review and meta-analysis are summarized in Table 2 and Table 3, respectively. All studies were judged to have a low risk of bias (Table 4). The meta-analysis demonstrated that metformin use is associated with decreased odds of VS growth when assessed as a binary outcome (pooled OR=0.45, 95% CI=0.29-0.71). These are graphically summarized in Figure 3. Studies demonstrated low heterogeneity and variance (I²=0%). A power analysis demonstrated that a total sample size of 220 patients (110 per study group) would be required to prospectively identify a true difference with approximately 80% power with an OR of 0.46.

Table 2.

Characteristics of included studies in the systematic review and meta-analysis.

Authors	Publication year	Study design	Measurement of growth	Statistical analysis	Result of interest	Level of Evidence (1a-5) †
Feng et al.	2020	Retrospective cohort	Linear; >25% increase in sum of perpendicular diameter products	Multivariable logistic regression model for growth as binary outcome	Metformin use is associated with less growth (OR=0.39, p=0.02)	3
Tran et al.	2021	Retrospective cohort	Volumetric; >20% increase in volume	Multivariable logistic regression model for growth as binary outcome	Metformin use is associated with less growth (OR=0.50, p=0.036)	3
Armstrong et al.	2021	Retrospective cohort	Volumetric; >20% increase in volume	Survival free of growth using Cox proportional hazards regression model	Metformin use is not associated with time to growth (HR=0.75, p=0.37)	3

^†According to Oxford Center for Evidence-Based Medicine criteria

OR: odds ratio, HR: hazard ratio

Table 3.

Demographics and tumor characteristics of included studies in the systematic review and meta-analysis.

Authors	Feng et al.	Tran et al.	Armstrong et al.
Total patients	149	387	361
Patients taking metformin (%)	42 (28)	46 (12)	19 (5)
Mean age (years)	69.6	60.6	ND
Male (%)	80 (54)	180 (47)	ND
Caucasian (%)	81 (54)	ND	ND
Mean length of observation (months)	30.6	30.8	ND
Mean initial tumor size (mm)	8	11.9	ND
Tumor growth (%)	65 (44)	216 (56)	319 (88)

Table 4.

Risk of bias assessment for the studies in the systematic review and meta-analysis.

	Feng et al.	Tran et al.	Armstrong et al.
Were the two groups similar and recruited from the same population?	Yes	Yes	Yes
Were the exposures measured similarly to assign people to both exposed and unexposed groups?	Yes	Yes	Yes
Was the exposure measured in a valid and reliable way?	Yes	Yes	Yes
Were confounding factors identified?	Yes	Yes	No
Were strategies to deal with confounding factors stated?	Yes	No	NA
Were the participants free of the outcome at the start of the study (or at the moment of exposure)?	Unclear	Unclear	Yes
Were the outcomes measured in a valid and reliable way?	Yes	Yes	Yes
Was the follow up time reported and sufficient to be long enough for outcomes to occur?	Yes	Yes	Yes
Was follow up complete, and if not, were the reasons to loss to follow up described and explored?	Yes	Yes	Yes
Were strategies to address incomplete follow up utilized?	NA	NA	NA
Was appropriate statistical analysis used?	Yes	Yes	Yes
Overall assessment	Low	Low	Low

NA: not applicable

Figure 3.

Forest plot of available cohort studies evaluating the effect of metformin on vestibular schwannoma growth utilized in the meta-analysis. Individual and pooled odds ratios and confidence intervals are displayed.

Discussion

The study herein sought to better understand the relationship between metformin and VS growth by reporting another institution’s experience and meta-analyzing the relevant literature. Our institution’s experience did not identify a statistically significant association between metformin on VS growth or time to VS growth in survival analysis. However, the meta-analysis did demonstrate a statistically significant reduction in the odds of VS growth with metformin use.

The initial investigation of metformin and VS likely stemmed from studies evaluating metformin’s effect on cancer incidence in the diabetic population²². A recent systematic review found that patients taking metformin had a statistically significant lower risk of being diagnosed with liver, colorectal, stomach, pancreas, and esophageal cancer, as well as cancer-related mortality²³. Specifically, exposure to metformin was associated with a 31% reduction in the risk of any cancer, and a 35% reduction in the risk of cancer mortality²³. The proposed antitumor mechanism of action is multifactorial. Directly, metformin activates adenosine monophosphate-activated protein kinase (AMPK) which inhibits mammalian target of rapamycin (mTOR), thereby promoting autophagy and apoptosis. It also downregulates c-myc proto-oncogene, which reduces tumor growth through anti-inflammatory and antiangiogenic effects^24–26. Metformin also inhibits oxidative phosphorylation in mitochondria, which creates reactive oxygen species and contributes to apoptosis²⁷. Indirectly, maintenance of a more euglycemic state with metformin may also contribute to its antitumor properties as hyperglycemia and hyperinsulinemia have been linked to tumor growth and proliferation^28–30. While these theories are promising, the true effects of each are unknown.

This study identified that 49% of all patients exhibited VS growth during follow up with a mean of 3.5 years of observation. This finding is in agreement with both prior natural history studies^17,31, as well as the three studies evaluating VS growth with metformin^14–16. Feng et al. was the first study to investigate metformin¹⁴. They evaluated 149 patients, 28% of which were taking metformin, and evaluated growth using linear measurements. They concluded that metformin users are significantly less likely to present with tumor growth at final follow-up on multivariable analysis (OR=0.39, p=0.02). Tran et al. was the largest study with 387 patients, 12% of which were metformin users¹⁵. They utilized volumetric measurements to assess for growth and found again that metformin use is significantly associated with reduced VS growth on multivariable analysis (OR=0.5, p=0.036). Finally, Armstrong et al. evaluated volumetric growth in 361 patients, 5% of which were metformin users¹⁶. In contrast to the prior two studies, they evaluated metformin’s effects in a time to event model using hazard regression models. They did not find a statistically significant association between metformin use and time to growth on multivariable analysis (HR=0.75, p=0.37). Finally, while the current study has the smallest cohort of all studies, it has the second largest proportion of metformin users at 17%. By evaluating survival free of tumor growth, we were able to evaluate metformin’s effects more analytically beyond assessing growth as a binary outcome, similar to Armstrong et al. However, by also employing logistic regression we were able to perform a meta-analysis in conjunction with Feng et al. and Tran et al. to provide statistical power.

Given the continued discrepancies between study outcomes, and even the outcomes presented herein, we are unable to argue causality between metformin use and VS growth. A true difference between metformin users and non-users would best be identified in a prospective trial, which would require 110 patients in each group based on the pooled OR found herein. While at this time we would not recommend prescribing metformin to non-diabetic patients with VS under active observation given unclear efficacy and potential metabolic side effects, it seems reasonable, however, to refer diabetic patients with VS under observation, particularly those who are not optimally controlled, to their Endocrinologist for consideration of metformin initiation.

While this study enhanced our understanding and has the highest level of evidence, to date, on the potential impact of metformin on VS growth, it has its limitations. First, as it is retrospective in nature it is subject to selection bias as some patients in either group may have been missed upon chart review. Secondly, this design precluded guaranteed medication compliance during observation, despite necessitating documented metformin use at first visit. Third, the use of linear measurements has been shown to lack sensitivity in measuring tumor growth when compared to volumetric measurements, which may have underestimated the rate of growth in both groups⁵. Fourth, reporting bias may exist in the systematic review as those studies which demonstrate a beneficial effect of metformin are more likely to be reported. Finally, the number of studies available for meta-analysis was limited based on available literature. With additional studies, a more precise result could be obtained. Ultimately, this information suggests that metformin remains a promising systemic therapy for VS patients and a randomized, placebo-controlled, prospective study is needed to identify a true effect.

Conclusion

There is conflicting evidence regarding the benefit metformin provides in inhibiting VS growth. Survival analysis does not demonstrate an association between the two, but a meta-analysis evaluating growth as a binary outcome suggests that metformin use is associated with a reduced odds of VS growth. An appropriately powered, randomized, prospective study is needed to identify a true difference between groups.