Understanding Permeability Changes in Vestibular Schwannomas as Part of the Dynamic Response to Radiosurgery Using Golden-Angle Radial Sparse Parallel Imaging: A Retrospective Study

Abstract

BACKGROUND AND OBJECTIVES:

Vestibular schwannomas demonstrate different responses after stereotactic radiosurgery (SRS), commonly including a transient loss of internal enhancement on postcontrast T1-weighted MRI thought to be due to an early reduction in tumor vascularity. We used dynamic contrast-enhanced based golden-angle radial sparse parallel (GRASP) MRI to characterize the vascular permeability changes underlying this phenomenon, with correlations to long-term tumor regression.

METHODS:

Consecutive patients with vestibular schwannoma who underwent SRS between 2017 and 2019, had a transient loss of enhancement after SRS, and had long-term longitudinal GRASP studies (6, 18, and 30 months) were included in this retrospective cohort analysis (n = 19). Using GRAVIS (https://gravis-imaging.org/gravis/), an analysis pipeline for GRASP studies, we extracted the key parameters normalized to the venous sinus from a region of interest within the tumor.

RESULTS:

The peak, area under the curve (AUC), and wash-in phase slope were significantly reduced at 6, 18, and 30 months after SRS (corrected P < .05), even while the internal enhancement returned in the tumors. Larger pre-SRS tumors were more likely to have a greater reduction in peak (P = .013) and AUC (P = .029) at 6 months. In a subset of patients (N = 13) with long-term follow-up, the median percentage reduction in tumor volume was 58% at a median of 62 months. These patients showed a strong correlation between peak, AUC, and wash-in phase slope changes at 6 months and tumor volume at the last follow-up.

CONCLUSION:

After SRS and loss of internal contrast uptake within vestibular schwannomas, a slow vascular permeability dynamic persisted, suggesting the presence of postradiation processes such as fibrosis. We show for the first time, using GRASP, a quantitative assessment of the vascular radiobiological effect.

ABBREVIATIONS:

DCE

dynamic contrast-enhanced

GR

Gardner-Robertson

GRASP

golden-angle radial sparse parallel

SRS

stereotactic radiosurgery

SSS

superior sagittal sinus

VS

vestibular schwannoma.

Vestibular schwannoma (VS) arises from the Schwann cell sheath of the vestibulocochlear nerve. It is a slow-growing tumor that commonly presents with tinnitus, sensorineural hearing loss, and disequilibrium. There is mounting evidence to link tumor growth to vascular permeability, neovascularization, and inflammatory cell infiltration.^1,2 The management options for VS include microsurgical resection and stereotactic radiosurgery (SRS), depending on clinical status and tumor morphology. The advantages of SRS are excellent tumor control (90%-99% at 5 years) with a high safety profile for facial nerve preservation and 41% to 79% hearing preservation at 5 years.^3,4

VSs demonstrate heterogeneous responses after SRS. MRI has been useful in noninvasively characterizing the structural changes within tumors. An early (6 months) loss of central signal enhancement has been described in 65% to 87% of VSs after SRS on postcontrast T1-weighted MRI, which is a delayed acquisition after contrast injection.^5-7 The signal enhancement typically returns at 12 to 24 months, resulting in a tumor contrast profile on postcontrast T1-weighted MRI similar to the pre-SRS study. One theory is that this transient change represents an early reduction in tumor vascularity, which may herald long-term growth control.

Functional or dynamic MRIs can better interrogate the underlying mechanisms of this dynamic response. Golden-angle radial sparse parallel (GRASP) MRI is a modern dynamic contrast-enhanced (DCE) sequence that combines radial k-space sampling with parallel imaging and compressed sensing for the acquisition of dynamic images with isotropic resolution, which can be reconstructed with high temporal resolution.^8-12 Moreover, due to the radial k-space sampling scheme, the contrast material can be injected during continuous acquisition (typically 20 seconds after the sequence starts), eliminating any risk of timing inaccuracies and significantly simplifying the clinical workflow. GRASP can cover large field-of-views at high spatial resolution and is inherently robust against motion artifacts. The temporal signal profile of lesions obtained with GRASP MRI is promising for localizing pituitary microadenomas and distinguishing progression from radiation necrosis in brain metastases after SRS.^8-12

We hypothesize that even after the internal contrast enhancement is re-established within a VS after SRS, its vascular permeability profile is permanently altered from baseline. In this study, we characterize the permeability changes corresponding to this dynamic pattern in cases with a positive response after SRS using a clinically acquired GRASP DCE-MRI database. We further correlate permeability changes to tumor volumes. In-house software, GRAVIS, was developed specifically for the fast analysis and visualization of GRASP 4D data sets, which will further improve clinical adoption.

METHODS

Patient Selection

From a cohort of patients with sporadic VS only who underwent SRS for VS between 2017 and 2019, we included only cases that demonstrated the central loss of contrast enhancement and subsequently regained a homogeneous appearance on postcontrast T1-weighted MRI. GRASP examinations must be available before SRS and at least once before 30 months post-SRS. We excluded patients who had prior surgery within 1 year of the baseline GRASP study or had confirmed tumor progression at the last follow-up (Supplemental Digital Content 1, http://links.lww.com/NEU/E595). We excluded the failed cases because (1) our study objective was to characterize a specific vascular radiobiological effect of SRS, and (2) the low number of failed cases limits any statistical comparison while at the same time introducing a confounding factor to the study objective. Hearing function was assessed using the Gardner-Robertson (GR) classification (Supplemental Digital Content 2, http://links.lww.com/NEU/E596). The Institutional Review Board approved this study. As a retrospective study with minimal risk, a waiver of authorization and informed consent was approved.

Radiosurgery Technique

SRS was performed using the Leksell Gamma Knife^® Icon^TM (Elekta) with frame application. Tumor, cochlea, and trigeminal nerve were identified and contoured on volumetric contrast-enhanced T1-weighted and constructive interference in steady-state MRI scans. Dose plans were created to ensure optimal coverage for conformal and selective irradiation within the constraints of neighboring structures at risk. Commonly, 12 to 13 Gy was prescribed to the 50% isodose line. We used 4.2 Gy as dose constraint (mean dose) for the ipsilateral cochlea unless the patient has GR class V hearing. Furthermore, we used sector blocking to reduce the mean cochlea dose as much as possible without compromising tumor coverage and minimal dose.

After SRS, the clinical visits and MRI studies occurred at 6 months, 18 months (ie, 1-year interval), and 30 months (ie, 1-year interval). Patients who decided to have MRIs at our institution had GRASP studies as per our standard protocol.

GRASP Acquisition

Dynamic GRASP studies were acquired before SRS, as well as at 6, 18, and approximately 30 months after the procedure using MAGNETOM 1.5 T or 3 T MRI systems (Siemens Healthcare). The technical acquisition details have been described previously.¹³ GRASP is a fat-suppressed 3D gradient-echo sequence with continuous radial stack-of-stars k-space acquisition according to the golden-angle scheme.¹⁴ Postacquisition, the data can be reconstructed with a selected temporal resolution by combining a variable number of radial k-space views into each temporal image frame, which enables retrospective customization of the desired temporal resolution. Parallel-imaging and compressed-sensing principles are used during the GRASP reconstruction to achieve high spatiotemporal resolution without compromising on image quality. A single bolus of 0.1 mL/kg gadobutrol (Gadavist, Bayer Schering Pharma) was administered intravenously at a rate of 4 mL/s after a 20-second delay from the start of the sequence. Imaging data were acquired continuously for 3 to 5 minutes. All GRASP studies were reconstructed by grouping 16 radial views per image frame, resulting in a median temporal resolution of 5.81 (range 5.08-8.06) seconds per frame.

Image Analysis

GRASP studies were analyzed using GRAVIS, an in-house software to facilitate fast analysis and visualization of multidimensional GRASP studies with large sizes, as required for viable clinical application of the technique. GRAVIS (https://gravis-imaging.org) for noncommercial and nondiagnostic use, provides a web-based user interface with various tools for the analysis of dynamic data, such as definition and comparison of regions of interest (ROIs), display of signal-time curves, and calculation of different quantitative metrics. Because computationally demanding operations are precalculated by an automatic preprocessing pipeline, studies load in GRAVIS within less than 5 seconds, regardless of the study size. The preprocessing pipeline designed by the VS studies includes automatic masking of the head, identification of the superior sagittal sinus (SSS), and normalization of the GRASP signal intensity relative to the signal within the SSS. Subsequently, quantitative maps are calculated for the following: peak, the area under the curve (AUC), and slopes during the wash-in and wash-out phases of the SSS.

To analyze each VS study, an ROI was placed in axial orientation within the area of internal contrast loss (Figure 1), ensuring the same spatial position across longitudinal studies. To correlate the GRASP parameters with long-term tumor control, we calculated the relative change in tumor volume at the last follow-up examination relative to baseline. Tumor volume was calculated using a contouring tool in the SRS planning software.

FIGURE 1.

A, vestibular schwannoma; B, demonstrating the central loss of contrast uptake at 6 months; C, filling in with contrast-enhancing material at 18 months after stereotactic radiosurgery; D, filling in with contrast-enhancing material at 30 months after stereotactic radiosurgery.

Statistical Testing

Clinical and radiographic variables were represented with descriptive statistics. The number of missing studies per time point is listed in Supplemental Digital Content 3 (http://links.lww.com/NEU/E597). The reasons for missing scans were missed follow-up, follow-up images acquired at an external institution, and raw data of acquired GRASP studies were not stored. Missing values were not imputed. A linear mixed-effects model was used to analyze longitudinal data, using a P-value of .05 with Tukey correction for multiple comparisons to reject the null hypothesis. Pearson correlation was used to assess the relationship between extracted GRASP parameters and tumor volume.

RESULTS

Clinical Follow-Up

Nineteen patients (mean age 58.2 ± 11.9 years, female-to-male ratio of 10:9) fitting our inclusion and exclusion criteria were analyzed for this study (Supplemental Digital Content 1, http://links.lww.com/NEU/E595). The distribution of Koos grade is listed in Table, with 13 patients (68%) having GR class I-II hearing preprocedure. The median tumor volume was 0.63 cm³. The median prescription dose was 12.5 Gy. The median follow-up time was 56.4 months. All patients in this study had tumor control at the last follow-up. Of the 13 patients with GR class I-II hearing, 6 remained in the same category, 4 had GR class III hearing, and 3 had missing data at their last follow-up.

TABLE.

Patient Demographics

Variable	Value
Age (y, median, range)	59 (35-84)
Sex
Men	9
Women	10
Koos grade
I	4
II	5
III	5
IV	5
Baseline GR hearing class (N)
I	9
II	4
III	2
IV	1
V	2
Volume (cm3, median, range)	0.63 (0.04-7.60)
Margin dose (Gy, median, range)	12.5 (12-13)

GR, Gardner-Robertson.

GRASP Parameters

All patients had a reduction in contrast enhancement in the core of the tumor on a 6-month follow-up scan from SRS and a return of the contrast uptake at 18 months on gadolinium-enhanced T1-weighted MRI (Figure 1). The GRASP time series from within the tumor at 6 months post-SRS were visibly different from baseline and remained altered at approximately 30 months post-SRS.

The tumor dynamic was evaluated with GRASP at baseline, 6 months, 18 months, and approximately 30 months after SRS. The extracted GRASP parameters, peak, AUC, wash-in phase slope, and wash-out phase slope are plotted for these time points in Figure 2. At 6 months, all parameters were significantly reduced relative to baseline. The peak and AUC values were at 46% of the baseline (corrected P < .001) at 6 months and remained decreased at 73% and 67% of the baseline (corrected P < .05, Figure 2A and 2B), respectively, at 30 months. At 30 months, these values remained significantly decreased.

FIGURE 2.

Boxplots of key parameters of the tumor golden-angle radial sparse parallel curve over time after radiosurgery. A, The AUC, B, peak, and slopes during the C, wash-in and D, wash-out phases of superior sagittal sinus were significantly altered from values before stereotactic radiosurgery by the linear mixed-effects model. **Corrected P < .001, *corrected P < .05. AU, arbitrary units; AUC, area under the curve.

The wash-in phase slope was at 54% of baseline at 6 months (corrected P < .001) and continued to be reduced at 71% (corrected P < .05) and 33% (corrected P < .001) of baseline at 18 and 30 months (Figure 2C). The wash-out phase slope was reduced at 6 months (corrected P = .05), but it was not significantly different from the baseline at 18 or 30 months (Figure 2D).

Correlations With Tumor Volume

Larger tumors are potentially more likely to demonstrate early internal loss of contrast uptake.⁷ We found that baseline tumor volume shared a strong correlation with the degree of reduction at 6 months in peak (r = −0.60, P = .013) and AUC (r = −0.54, P = .029, Supplemental Digital Content 4, http://links.lww.com/NEU/E598).The linear correlation between baseline tumor volume and change in wash-in and wash-out phase slopes was not significant (r = −0.48, P = .061; r = −0.30, P = .27).

To test whether these early 6-month changes in vascular permeability can also predict long-term tumor control, we correlated the GRASP parameters to the relative reduction in tumor volume at the last follow-up. Because VSs are known to have variable changes in volume after SRS, such as pseudoprogression, we excluded patients who did not have a follow-up longer than 4 years. Previous studies show delayed regressions continue and appear to plateau at approximately 4 years.^15-17 In doing so, 13 patients were included in this analysis. There was a median of 58% (range 7.6%-89%) reduction in tumor volume at a median follow-up time of 62 (range 50-84) months.

In representing the data, we simplified the relative change to the percentage reduction in the tumor volume and parameters. A negative change indicated an increase in value. The reduction in peak (r = 0.66, P = .015), AUC (r = 0.66, P = .015), and wash-in slope (r = 0.74, P = .0041) at 6 months strongly predicted the reduction in tumor volume at the last follow-up (Figure 3). The positive correlation is interpreted as a greater reduction in the peak, which was associated with greater long-term tumor regression after SRS. Changes in wash-out slope or parameters at 18 months had no significant correlation.

FIGURE 3.

Percentage reduction in tumor volume plotted against percentage reduction A, peak, B, AUC, C, wash-in phase slope, and D, wash-out phase slope at 6 months after stereotactic radiosurgery. Pearson correlations and P-values are indicated for respective relationships. Negative positions indicate an increase in value relative to the baseline. AUC, area under the curve.

DISCUSSION

This study used GRASP images to characterize the permeability changes corresponding to a dynamic contrast pattern in VS that had a positive response after SRS. We show evidence to support our hypothesis that the vascular permeability within VS is slowed and reduced after SRS, corresponding to an internal loss of contrast, and it remains permanently reduced even after contrast enhancement is re-established. Specifically, significant changes were seen in the signal peak, AUC, and wash-in slope. Based on previous studies that showed a positive association between DCE model parameter K^trans and faster tumor growth, greater inflammatory cell infiltration, and tumor progression,^1,2,18-20 the permanently reduced permeability profile after SRS suggests an internal development of scar tissue may account for the return in contrast enhancement (and not tumor regrowth). We also showed that greater reductions in these dynamics predicted greater long-term tumor regression. Cases that did not respond to SRS are the subject of a separate and future study.

Previously, model-based parameters from DCE-MRIs have been useful in understanding the perfusion and permeability changes in tumors and predicting tumor response after SRS using K^trnas.^1,2,18-20 The disadvantage of these DCE-MRI model-derived parameters is their complexity. Our study used clinical GRASP studies, which offer several advantages over conventional DCE-MRIs, including inherent fat-suppression, inherent robustness to patient motion, a simple and fast acquisition protocol, and high temporal-spatial resolution, leading to improved usability and accessibility in the clinic.^21-23 With a simple acquisition protocol, the GRASP method does not rely on a precise timing of contrast injection and start of the dynamic sequences. Moreover, the higher reproducibility across different software implementations is an advantage of the model-free analysis using descriptive parameters such as peak, AUC, and the wash-in and wash-out phase slopes. In this study, we used the GRAVIS software, a clinically focused, streamlined analysis pipeline and interface for GRASP studies, which has been specifically designed to eliminate typical workflow bottlenecks in clinical applications and to enable wider implementation of the GRASP DCE-MRI approach. GRASP could be included with structural MRI features as part of a radiomics-based machine-learning approach to predict tumor behavior and response to treatment.^7,24

An early transient loss of central contrast enhancement has been described in 65% to 87% of VSs after SRS.^25-28 The underlying mechanism is poorly understood but theorized to include tumor necrosis and vascular damage.²⁹ Vascular damage from SRS reduces blood perfusion, leading to indirect tumor cell death in addition to direct tumor DNA damage.^30,31 The subsequent return of contrast enhancement could be explained by scar formation, tumor regrowth, or inflammatory infiltration. Dynamic MRI can help differentiate between these processes. Previous studies with DCE-MRI and pathology confirmation show persistently delayed enhancement in the arterial phase can differentiate scar and granulation tissue from recurrent tumor in various pathologies.^32,33 In another study of gliomas, the maximum slope of enhancement was the best single predictor of tumor vs treatment-related changes.³⁴ Therefore, our interpretation of the data is that after the initial internal injury to cells and blood vessels, scars are laid down within the tumor.

Although this phenomenon is often associated with slight volume expansion, it is thought to be an independent process from pseudoprogression, another response pattern after SRS consisting of transient volume expansion. The association and timing of these 2 patterns have been inconsistently found in several morphological studies.^5-7,27,35,36 The relationship between GRASP parameters and pseudoprogression or progression is the subject of a separate study.

Furthermore, because the loss of internal contrast is thought to be due to cellular necrosis and vascular obliteration, it would follow this phenomenon and result in favorable long-term tumor control and regression.^25,29 However, this relationship has been debated.²⁷ Data from our study support the prior hypothesis, ie, in patients with internal loss of contrast, greater reduction in vascular permeability does predict greater tumor progression long term. This correlation analysis requires further validation in a broader patient population, along with the analysis of confounders and other explanatory variables such as baseline tumor volume and long-term tumor progression.

Limitations

Our study characterized the changes in vascular permeability in a selected group of patients and lacked a control group without internal contrast uptake or tumors that progressed after SRS. As such, our results should not be interpreted as a predictor of SRS failure. Further investigations are needed to compare the permeability values between tumors with long-term regression, stability, or progression after SRS. Another limitation is the small sample size despite a much larger cohort of patients who undergo SRS in the inclusion period. For instance, excluding patients who had baseline MRIs performed at an external institution and thus did not have a baseline GRASP study significantly restricted the sample size. On the other hand, we excluded patients who had microsurgery within 1 year of the baseline study as a loss in enhancement can be found in nearly one-half of the residual tumors within that time period.³⁷ Overall, the limited case numbers can introduce bias and reduce the robustness of our results, particularly the correlation analysis between changes in parameters and tumor volume, which will require further validation.

Finally, another limitation of our methodology is the placement of the ROI in an unblinded fashion. We considered the alternative of extracting the GRASP signal from the full tumor segmentation. However, because the goal was to characterize the changes within the internal area of the tumor, demonstrating the response of interest, we felt that this was an appropriate approach.

CONCLUSION

We found the vascular permeability within the tumor was permanently slowed, even as the irradiated tumor internally regains contrast enhancement. This dynamic change, supported by previous literature, is reflective of posttreatment change rather than tumor recurrence. We also showed a greater reduction of the GRASP parameters correlated with greater long-term tumor regression. Clinically acquired GRASP sequences allowed a quantitative assessment of vascular radiobiological response in tumor tissue after SRS.

Funding

This study did not receive any funding or financial support.

Disclosures

Douglas Kondziolka receives research support from Brainlab AG. The other authors have no personal, financial, or institutional interests in any of the drugs, materials, or devices described in this article. Kenneth Bernstein receives financial support from Neuropoint Alliance.

SUPPLEMENTAL DIGITAL CONTENT

Supplemental Digital Content 1. Figure S1. Flowchart of patients selected for GRASP analysis.

Supplemental Digital Content 2. Table S1. Gardner-Roberson Hearing Classification System. Used with permission by Silverstein, H., McDaniel, A. and Norrell, H. (1986), Hearing preservation after acoustic neuroma surgery with intraoperative direct eighth cranial nerve monitoring: Part II. A classification of results. Otolaryngology-Head and Neck Surgery, 95: 285-291. Wiley https://doi.org/10.1177/01945998860953P104.

Supplemental Digital Content 3. Table S2. Distribution of missing scans for patients across time points.

Supplemental Digital Content 4. Figure S2. Plots of correlation between changes in GRASP parameters (A) peak, (B) AUC, (C) slope during wash-in phase of the SSS, and (D) slope during wash-out phase of the SSS, 6 months after SRS to baseline tumor volume.

COMMENTS

The study investigates the innovative dynamic contrast-enhanced (DCE) GRASP MRI technique for assessing vascular permeability in vestibular schwannoma (VS) following stereotactic radiosurgery (SRS). This research is particularly significant as it explores the long-term effects of SRS on tumor vasculature, an area that has received limited attention in existing literature. The findings indicate that, despite an initial temporary loss of contrast enhancement, the vascular permeability in VS is altered over the long term. This is evidenced by marked reductions in key parameters such as peak enhancement, area under the curve (AUC), and wash-in slope metrics, which are essential for understanding tumor perfusion dynamics.

A noteworthy aspect of the study is the observed correlation between early changes in these vascular metrics and subsequent tumor volume reduction. This relationship suggests that GRASP MRI may serve as a valuable predictive tool for anticipating long-term tumor control following SRS. The ability to monitor vascular changes noninvasively positions GRASP MRI as an essential adjunct in the clinical management of vestibular schwannomas, potentially guiding treatment decisions and improving patient outcomes.

Moreover, the application of GRASP MRI in clinical practice could enhance our understanding of tumor biology and behavior, providing insights into the mechanisms underlying treatment responses. This could lead to more personalized approaches in managing vestibular schwannomas, where treatment strategies could be tailored based on the dynamic vascular characteristics of individual tumors.

However, the study does highlight the need for further research to validate these findings in larger cohorts, which would strengthen the evidence base for the routine use of GRASP MRI in clinical settings. Future studies could also explore the potential of integrating GRASP MRI findings with other imaging modalities to provide a more comprehensive assessment of tumor characteristics and treatment effects.

In conclusion, this study underscores the promise of GRASP MRI as a noninvasive tool that not only enhances our understanding of tumor behavior but also holds the potential to improve clinical outcomes in the management of vestibular schwannomas.

Juan Manuel Altamirano

Ciudad de México, Mexico