Abstract
Purpose
To determine factors associated with increased risk of finding new and/or enlarged brain metastases (BM) on GammaKnife™ (GK) MRI and their impact on patient outcomes.
Results
43.9% of patients showed BM growth, 32.9% had additional brain metastases (aBM), and 18.1 % had both. Initial brain metastasis velocity (iBMV) was associated with finding aBM. Time between diagnostic MRI (dMRI) and GK MRI was associated with interval growth and each day increased this risk by 2%. Prior brain metastasectomy and greater time between either dMRI or latest extracranial RT and GK MRI predicted both aBM and BM growth. aBM and/or BM growth led to management change in 1.8% of cases and were not associated with OS or incidence of distant intracranial failure.
Conclusions
Number of metastases seen on dMRI and iBMV predicted both aBM and/or BM growth, however, these factors did not significantly affect survival or incidence of distant intracranial failure.
Keywords: GammaKnife, SRS, brain metastases
INTRODUCTION
The most common cause of intracranial tumors in adults is metastatic cancer, which accounts for more than half of intraparenchymal brain disease and occurs in 20-40% of patients with cancer [1]. Improvements in survival from modern systemic therapy, increased utilization of brain MRI, and increased rate of diagnoses of cancer in the elderly has contributed to an increase in the incidence of diagnosis of brain metastases (BM), however, prognosis for many of these patients remains poor [2-4].
The management of BM has become increasingly complex and often the specific number of BM impacts the choice of treatment [3, 5-9]. Stereotactic radiosurgery (SRS) is the current treatment of choice of a limited number of lesions (i.e., 1-4) and in the salvage setting after whole-brain radiotherapy (WBRT) [8, 10-12]. However, prospective evidence has shown that SRS for 5-10 brain metastases does not portend worse overall survival compared to 2-4 lesions and as such, use of SRS for a greater number of lesions has become more commonplace [3, 13-16].
Most institutions, including ours, currently employ the use of same-day MRI for GK planning and treatment. Previous studies have investigated factors that may predict additional brain metastases (aBM) on GK MRI. Subsequent studies have shown that there is a greater risk of finding aBM if multiple BM were seen on dMRI, those with aBM had significantly worse median overall survival compared to those without, and that the risk of finding aBM was inversely proportional to the size of the index brain metastasis [17, 18]. The presence of aBM has been shown to predict a higher incidence of distant intracranial failure, although this did not translate into a difference in survival [17]. The lack of differences in patient outcomes in these studies for patients with a greater number of brain lesions may be partly explained by the fact that total lesion number is only part of the picture of total intracranial disease burden. Recent evidence has shown that cumulative lesion volume is also an important factor to consider in patients with BM and is at least as important a predictor of outcomes after SRS, if not more so, than total lesion number alone [19-22]. However, these studies, however, did not investigate the impact of metastasis growth on outcomes in these patients. We therefore sought to investigate factors associated with finding aBM on GK MRI and/or growth of previously identified lesions and correlate these findings with patient outcomes.
Here, we report our experience with additional findings on high-resolution, same day, thin-slice, gadolinium-enhanced MRI for GK treatment planning as compared to the diagnostic MRI that led to that treatment. To our knowledge, this is the first study to determine factors associated with an increased risk of finding new and/or enlarged brain metastases on GK MRI compared to dMRI. We also sought to correlate these findings to patient outcomes.
MATERIALS AND METHODS
Study Population
This study is a retrospective, institutional review board-approved, single-institution review of a total of 377 consecutive, unique, patients (491 individual procedures) who were treated with GK at Indiana University between May 2006 and December 2018. All patients deemed appropriate for GK were included in this study, including those who previously underwent whole brain radiation. All but one patient received same-day, gadolinium-based contrast per manufacturers recommendations (Multihance® (Bracco), Prohance® (Bracco), or Magnevist® (Bayer)) for GK MRI for comparison to dMRI. The other patient underwent CT-based SRS planning due to MRI contraindication. The institutional protocol for those SRS planning scans are to use 1mm slice thickness with 3D inversion recovery spoiled gradient echo (IR-SPGR) sequences. Patient baseline characteristics were recorded from the radiation oncologists’ consultation notes, referring providers’ notes, and the electronic medical record, including age, sex, presence/absence of extracranial metastases, prior imaging parameters, presence/absence of neurological symptoms, prior WBRT, use/timing of systemic therapy (chemotherapy and/or immunotherapy) in relation to GK, prior resection of brain metastasis, prior use of SRS, and histology. Initial brain metastasis velocity (iBMV) was defined as the number of brain metastases at initial GK divided by the time in years from diagnosis as in previous reports [23, 24].
Image Analysis and GK Procedure
Thin-slice, gadolinium-enhanced GK planning MRI was performed after placement of a stereotactic frame (Leksell® Coordinate Frame G), which was fixed in place under local anesthetic by neurosurgery via four pin sites. The GK MRI images were then loaded into the treatment planning system (GammaPlan®) and each lesion was measured, then contoured individually. The final number of treated BM was determined using information gained from T1- and T2-weighted images in the axial, coronal, and sagittal planes upon agreement between the treating radiation oncologist, neuroradiologist, and neurosurgeon. Brain metastasis growth was defined as an increase in any lesion dimension (x, y, or z coordinates) per neuroradiology. Treatment planning focused on maximizing coverage and conformality and then performed in a single fraction after the plan was approved by a neurosurgeon, a radiation oncologist and the medical physics team.
Statistical Analysis
Continuous variables were summarized using median, minimum, and maximum and include time between last any infusion and GK (days), time between latest cranial radiation (RT) and GK (days), time between latest extracranial RT and GK (days), minimum time between from any RT to GK (days), time between dMRI and GK (days), difference between dMRI and GK MRI slice thickness (mm), and iBMV. Categorical variables were summarized using frequency and percentage and included primary site, histology, grade, and binary variables including triple negative receptor status if breast primary, symptoms, any systemic treatments prior to GK, any intracranial RT prior to GK, and any extracranial RT prior to GK. Chi-square tests were conducted to assess the relationship between categorical outcome variables and categorical covariates. Fisher’s exact test was used instead of chi-square if < 5 patients fit criteria for each specific analysis. Wilcoxon rank-sum tests were conducted to assess the relationship between categorical outcome variables and continuous covariates.
Univariable and multivariable logistic regression were used to assess the odds of finding aBM, interval growth of previously identified BM, or both, based on disease and treatment characteristics. Variables that were significant at p < 0.20 in the univariable analysis were included in a stepwise selection procedure to select which variables would be included in the final multivariable logistic regression model.
Overall survival (OS) was calculated in two different ways. The first method was calculating from the date of primary diagnosis until the date of death or last follow-up. The second method was from the first GK treatment until the date of death or last follow-up. Patients who were lost to follow-up or remain alive were censored at their last known follow-up. The OS probability was estimated by the Kaplan-Meier method and compared between groups using the log-rank test. Logistic regression analysis and Fine-Gray regression modeling were used to assess factors predictive of aBM and/or interval growth. Significance was defined as a two-sided p-value < 0.05. All analyses were performed using SAS version 9.4 (SAS Institute; Cary, NC).
RESULTS
Patient Baseline Characteristics
We identified 377 unique patients undergoing a total of 491 independent GK procedures between May 2006 to December 2018. The median patient age was 58 years at the time of GK (range 21-87 years) and the median time between dMRI and GK MRI of 18 days (range 0-98 days). At consultation, 49.1% of patients had neurological symptoms, and prior intracranial treatment with WBRT, surgical resection, or Linac-based SRS was found in 15.7%, 2.9%, and 7.2%, respectively. Systemic therapy was given prior to GK in 22.6% of patients and the last infusion was given at a median of 38 days prior to GK (range 1-887 days). Extracranial radiation was given in 21.0% of patients at a median of 208 days prior to GK (range 3-4498 days). The most common histologies were adenocarcinoma (31.3%), melanoma (21.0%), and breast cancer (17.2%). 80.7% of evaluable MRIs found that the slice thickness of Dx MRI and GK MRIs were equivalent (median slice thickness of 1 mm for both) and only 9.4% of patients had a Dx MRI with 5 mm slices (data not shown). Additional patient characteristics are summarized in Table 1.
Table 1.
Baseline patient characteristics for total cohort
| Characteristic | Value |
|---|---|
| Total number of patients | 377 |
| Patient age | |
| Median, years (range) | 58 (21–87) |
| <40 | 25 (6.6) |
| 40-49 | 61 (16.2) |
| 50-59 | 120 (31.8) |
| 60-69 | 115 (30.5) |
| ≥70 | 56 (14.9) |
| Sex | |
| Male | 160 (42.4) |
| Female | 217 (57.6) |
| Extracranial metastases | |
| Yes | 50 (13.3) |
| No | 327 (86.7) |
| Neurological symptoms | |
| Yes | 185 (49.1) |
| No | 192 (50.9) |
| Prior WBRT | |
| Yes | 59 (15.7) |
| No | 318 (84.4) |
| Systemic therapy prior to GK | |
| Yes | 85 (22.6) |
| No | 292 (77.5) |
| Prior brain metastasis resection | |
| Yes | 11 (2.9) |
| No | 366 (97.1) |
| Prior radiosurgery | |
| Yes | 27 (7.2) |
| No | 350 (92.8) |
| Histology | |
| Adenocarcinoma | 118 (31.3) |
| Breast | 65 (17.2) |
| Melanoma | 79 (21.0) |
| Neuroendocrine | 15 (4.0) |
| RCC | 37 (9.8) |
| SCC | 32 (8.5) |
| Other | 31 (8.2) |
Detection of aBM and/or BM Interval Growth on GK MRI
Overall, 374 of 377 total patients met criteria for evaluation for aBM, which was found in 32.9% of patients. On average, one aBM was found on GK MRI (mean 3, range 1-17) vs dMRI (mean 2, range 0-14). On univariate analysis (UVA) for factors significant for predicting aBM on GK MRI, patient age (per year), presence of extracranial metastases, recent receipt of systemic therapy, prior brain metastasectomy, prior WBRT, prior receipt of extracranial XRT, number of metastases on dMRI (per metastasis), and iBMV were significant (Table 2A), and were thus included in the multivariable analysis (MVA). On MVA, only the number of metastases on dMRI (per metastasis) (OR 1.19 p = 0.006), and iBMV (OR 1.01, p = 0.042) remained significant predictors of aBM (Table 2B). There was no association between length of time between dMRI and GK MRI (p = 0.930), presence of extracranial metastases (p = 0.614), dMRI slice thickness (p = 0.844), prior systemic therapies (p = 0.127) or prior intracranial surgery or radiation (p = 0.290), and discovery of aBM. An additional analysis was performed to determine if breast cancer receptor status influenced discovery of aBM on GK MRI, and no significant association was found (Supplemental Tables S1A and S1B).
Table 2A.
Univariable analysis for factors significant for predicting aBM on GK MRI
All variables with p < 0.20 in univariable analysis were entered into a stepwise selection procedure to determine the multivariable model
| Univariable Analysis | |||
|---|---|---|---|
| Characteristic | n | OR (95% CI) | p-value |
| Patient age (per year) | 374 | 0.99 (0.97 – 1.00) | 0.10* |
| Male sex | 374 | 1.10 (0.71 – 1.70) | 0.66 |
| KPS, 80+ vs <80 | 327 | 1.09 (0.66 – 1.78) | 0.74 |
| Extracranial metastases present | 374 | 0.53 (0.26 – 1.07) | 0.08* |
| Neurological symptoms present | 374 | 1.29 (0.84 – 1.99) | 0.24 |
| Systemic therapy at time of SRS | 374 | 1.47 (0.89 – 2.43) | 0.13* |
| Time between last any infusion and GK (per day) | 85 | 0.999 (0.998 – 1.001) | 0.61 |
| Prior brain metastases resection | 374 | 2.49 (0.75 – 8.33) | 0.14* |
| Prior WBRT delivered | 374 | 1.47 (0.83 – 2.61) | 0.18* |
| Prior SRS delivered | 374 | 0.69 (0.28 – 1.67) | 0.41 |
| Time between latest cranial XRT and GK (per day) | 85 | 1.00 (0.998 – 1.001) | 0.53 |
| Prior delivery of extracranial XRT | 374 | 0.68 (0.39 – 1.17) | 0.16* |
| Time between latest extracranial XRT and GK (per day) | 79 | 1.00 (0.999 – 1.001) | 0.95 |
| Minimum time between any XRT to GK (per day) | 134 | 1.00 (0.999 – 1.001) | 0.86 |
| Time between dMRI and GK MRI (per day) | 374 | 1.00 (0.98 – 1.01) | 0.59 |
| dMRI Slice Thickness | 338 | 1.03 (0.88 – 1.21) | 0.69 |
| Number of metastases on dMRI (per metastasis) | 373 | 1.21 (1.07 – 1.37) | 0.0020* |
| Histology Breast vs Adenocarcinoma Melanoma vs Adenocarcinoma Neuroendocrine vs Adenocarcinoma Other vs Adenocarcinoma RCC vs Adenocarcinoma SCC vs Adenocarcinoma |
374 | 1.84 (0.97 – 3.50) 1.93 (1.05 – 3.53) 1.34 (0.43 – 4.23) 1.28 (0.54 – 3.01) 0.86 (0.37 – 2.03) 1.15 (0.48 – 2.78) |
0.27 |
| iBMV | 368 | 1.01 (1.00 – 1.02) | 0.0340* |
*p < 0.20
Table 2B.
Multivariable analysis for factors significant for predicting aBM on GK MRI
| Multivariable Analysis | ||
|---|---|---|
| Characteristic | OR (95% CI) | p-value |
| Number of metastases on dMRI (per metastasis) | 1.19 (1.05 – 1.34) | 0.0055 |
| iBMV | 1.01 (1.00 – 1.02) | 0.0420 |
Table S1A.
Chi-square test for influence of breast cancer receptor status on discovery of aBMs on GK MRI.
| Chi-Square Tests | ||||
|---|---|---|---|---|
| Variable | n | Interval New Mets? | p-value | |
| No | Yes | |||
|
Total, n (%)
ER Status, n (%) |
70 | 41 (58.6) | 29 (41.4) | - |
| Positive Negative |
66 | 21 (53.9) 18 (46.2) |
12 (44.4) 15 (55.6) |
0.4526 |
| PR Status, n (%) | ||||
| Positive Negative |
66 | 13 (33.3) 26 (66.7) |
7 (25.9) 20 (74.1) |
0.5197 |
| HER Status, n (%) | ||||
| Positive Negative |
65 | 16 (42.1) 22 (57.9) |
13 (48.2) 14 (51.9) |
0.6291 |
| ER and HER Status, n (%) | ||||
| Positive Negative |
64 | 6 (23.1) 20 (76.9) |
10 (26.3) 28 (73.7) |
0.7688 |
| Triple Negative, n (%) | ||||
| No Yes |
70 | 30 (73.2) 11 (26.8) |
20 (69.0) 9 (31.0) |
0.7012 |
Table S1B.
Odds ratio for influence of breast cancer receptor status on discovery of aBMs on GK MRI.
| Odds Ratios | ||||
|---|---|---|---|---|
| Variable | n | Odds Ratio | 95% Confidence Interval | p-value |
| ER Status, Positive vs Negative | 66 | 0.69 | 0.26 – 1.84 | 0.4534 |
| PR Status, Positive vs Negative | 66 | 0.70 | 0.24 – 2.08 | 0.2776 |
| HER Status, Positive vs Negative | 65 | 1.28 | 0.47 – 3.44 | 0.6293 |
| ER and HER Status, Positive vs Negative | 64 | 0.84 | 0.26 – 2.69 | 0.2968 |
| Triple Negative, Yes vs No | 70 | 1.23 | 0.43 – 3.50 | 0.7014 |
Growth of BM initially found on dMRI was evaluable in 371 of the 377 patients and was found in 43.9% of patients, most frequently in patients with adenocarcinoma (30.1%). Time between dMRI and GK MRI was significantly associated with BM growth (p < 0.001) (Table 3A) and when measured as a continuous variable, iBMV was also significantly predictive of BM growth (p < 0.001) (Table 3B). On UVA, recent systemic treatment, time between dMRI and GK MRI, and number of metastases on dMRI (per metastasis) were found to be significant predictors of BM growth (Table 4A) and were included in the MVA. On MVA, time between dMRI and GK MRI (OR 1.02, 95% CI 1.01-1.04) and number of metastases on dMRI (per metastasis) (OR 1.19, 95% CI 1.05-1.34) remained significantly associated with findings of interval growth of previously identified BM (Table 4B). There was no association between histology, prior intracranial therapy, prior systemic therapy, or dMRI slice thickness on BM growth.
Table 3A.
Association between length of time between scans and BM growth on GK MRI
| Interval Growth | N | Median (Min – Max) | Mean ± SD |
|---|---|---|---|
| Yes | 163 | 20 days (2 – 98) | 25.7 days ± 17.78 |
| No | 208 | 16 days (1 - 93) | 20.1 days ± 15.03 |
Table 3B.
Descriptive statistics for continuous variables with respect to interval BM growth
| Continuous Variables | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Variable | Interval New Met Growth? | ||||||||
| No | Yes | ||||||||
| n | Median | Min | Max | n | Median | Min | Max | p-value | |
| Time between Last Any Infusion and GK (days) | 51 | 23 | 1 | 1,521 | 34 | 63 | 3 | 685 | 0.1783 |
| Time between Latest Cranial XRT and GK (days) | 54 | 253.5 | 32 | 1,605 | 31 | 245 | 26 | 1,163 | 0.7702 |
| Time between Latest Extracranial XRT and GK (days) | 58 | 279 | 4 | 2,926 | 21 | 151 | 3 | 4,998 | 0.1885 |
| Minimum Time from Any XRT to GK (days) | 89 | 200 | 4 | 2,926 | 45 | 183 | 3 | 4,998 | 0.3425 |
| Time between MRI and GK (days) | 250 | 18 | 1 | 98 | 124 | 18 | 2 | 94 | 0.9299 |
| Difference between Dx and GK MRI Slice | 224 | 0 | -0.3 | 5 | 113 | 0 | -0.3 | 5 | 0.7625 |
| iBMV (metastasis/year) | 245 | 1 | 0 | 182.6 | 123 | 2.7 | 0.1 | 243.5 | <.0001* |
*p = 0.0002
Table 4A.
Univariable analysis for factors significant for predicting interval BM growth on GK MRI
All variables with p < 0.20 in univariable analysis were entered into a stepwise selection procedure to determine the multivariable model
| Univariable Analysis | |||
|---|---|---|---|
| Characteristic | n | OR (95% CI) | p-value |
| Patient age (per year) | 371 | 0.99 (0.97 – 1.01) | 0.24 |
| Male sex | 371 | 0.82 (0.54 – 1.24) | 0.35 |
| KPS, 80+ vs <80 | 325 | 0.84 (0.53 – 1.34) | 0.47 |
| Extracranial metastases present | 371 | 0.91 (0.50 – 1.67) | 0.77 |
| Neurological symptoms present | 371 | 0.98 (0.65 – 1.47) | 0.91 |
| Systemic therapy at time of SRS | 371 | 0.67 (0.41 – 1.10) | 0.12* |
| Time between last any infusion and GK (per day) |
85 | 0.999 (0.997 – 1.001) | 0.49 |
| Prior brain metastases resection | 371 | 1.07 (0.32 – 3.56) | 0.92 |
| Prior WBRT delivered | 371 | 0.84 (0.47 – 1.49) | 0.55 |
| Prior SRS delivered | 371 | 1.20 (0.55 – 2.63) | 0.65 |
| Time between latest cranial XRT and GK (per day) |
83 | 1.000 (0.99 – 1.001) | 0.93 |
| Prior delivery of extracranial XRT | 371 | 0.75 (0.45 – 1.25) | 0.27 |
| Time between latest extracranial XRT and GK (per day) | 78 | 1.000 (0.999 – 1.000) | 0.49 |
| Minimum time between any XRT to GK (per day) | 132 | 1.000 (0.999 – 1.000) | 0.48 |
| Time between dMRI and GK MRI (per day) | 371 | 1.02 (1.01 – 1.04) | 0.0016* |
| dMRI Slice Thickness | 336 | 1.08 (0.93 – 1.26) | 0.30 |
| Number of metastases on dMRI (per metastasis) | 371 | 1.17 (1.03 – 1.32) | 0.0135* |
| Histology Breast vs Adenocarcinoma Melanoma vs Adenocarcinoma Neuroendocrine vs Adenocarcinoma Other vs Adenocarcinoma RCC vs Adenocarcinoma SCC vs Adenocarcinoma |
371 | 1.17 (0.63 – 2.16) 1.34 (0.75 – 2.38) 0.70 (0.23 – 2.19) 0.70 (0.30 – 1.64) 1.49 (0.71 – 3.12) 1.14 (0.51 – 2.59) |
0.69 |
| iBMV | 366 | 1.01 (1.00 – 1.02) | 0.25 |
*p < 0.20
Table 4B.
Multivariable analysis for factors significant for predicting interval BM growth on GK MRI
| Multivariable Analysis | ||
|---|---|---|
| Characteristic | OR (95% CI) | p-value |
| Time between dMRI and GK MRI (per day) | 1.02 (1.01 – 1.04) | 0.0010 |
| Number of metastases on dMRI (per metastasis) | 1.19 (1.05 – 1.34) | 0.0080 |
Evaluation for both aBM and interval growth was possible in 371 patients and found in 18.1% of patients. Time between dMRI and GK MRI (p = 0.04), time between latest extracranial XRT and GK (p = 0.07), and iBMV (p = .0001) were significantly associated with both aBM and growth of previously identified BM, when measured as continuous variables (Supplemental Table S2). There was also a trend toward histology impacting finding both aBM and BM growth (p = 0.06), with melanoma showing both 32.8% of the time, and 22.4% of the time in adenocarcinoma (data not shown). Prior intracranial/systemic therapies and dMRI slice thickness were not significantly associated with finding both aBM and interval growth of BM. On UVA, patient age (per year), KPS (≥ 80 vs < 80), prior brain metastasectomy, prior delivery of extracranial XRT, number of metastases on dMRI (per metastasis), and iBMV were found to be significant (Table 5A), with prior brain metastasectomy (OR 3.94, 95% CI 1.14-13.6) and number of metastases on dMRI (per metastasis) (OR 1.21, 95% CI 1.06-1.38) remaining a significant predictor of both on MVA (Table 5B).
Table S2.
Descriptive statistics for continuous variables with respect to finding both aBM and interval BM growth.
| Continuous Variables | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Both Interval New Mets and Met Growth? | |||||||||
| Variable | No | Yes | p-value | ||||||
| n | Median | Min | Max | n | Median | Min | Max | ||
| Time between Last Any Infusion and GK (days) | 69 | 27 | 1 | 1,521 | 16 | 63 | 3 | 307 | 0.3115 |
| Time between Latest Cranial XRT and GK (days) | 69 | 252 | 26 | 1,605 | 14 | 275.5 | 71 | 1,163 | 0.7750 |
| Time between Latest Extracranial XRT and GK (days) | 69 | 270 | 4 | 4,998 | 9 | 81 | 3 | 361 | 0.0068* |
| Minimum Time from Any XRT to GK (days) | 111 | 193 | 4 | 4,998 | 21 | 154 | 3 | 1,163 | 0.1882 |
| Time between MRI and GK (days) | 304 | 17 | 1 | 98 | 67 | 20 | 8 | 94 | 0.0396* |
| Difference between Dx and GK MRI Slice | 276 | 0 | -0.3 | 5 | 60 | 0 | -0.3 | 4 | 0.5189 |
| Initial Brain Met Velocity (mets/year) | 300 | 1.2 | 0 | 243.5 | 66 | 2.8 | 0.1 | 182.6 | 0.0013* |
Table 5A.
Univariable analysis for factors significant for predicting finding both aBM and interval BM growth on GK MRI
All variables with p < 0.20 in univariable analysis were entered into a stepwise selection procedure to determine the multivariable model
| Univariable Analysis | |||
|---|---|---|---|
| Characteristic | n | OR (95% CI) | p-value |
| Patient age (per year) | 371 | 0.99 (0.96 – 1.01) | 0.17* |
| Male sex | 371 | 0.77 (0.44 – 1.32) | 0.34 |
| KPS, 80+ vs <80 | 325 | 0.68 (0.38 – 1.21) | 0.19* |
| Extracranial metastases present | 371 | 0.58 (0.24 – 1.43) | 0.24 |
| Neurological symptoms present | 371 | 1.18 (0.70 – 2.01) | 0.53 |
| Systemic therapy at time of SRS | 371 | 1.07 (0.57 – 1.99) | 0.83 |
| Time between last any infusion and GK (per day) | 85 | 0.999 (0.996 – 1.002) | 0.55 |
| Prior brain metastases resection | 371 | 4.01 (1.19 – 13.5) | 0.0255* |
| Prior WBRT delivered | 371 | 1.10 (0.54 – 2.26) | 0.79 |
| Prior SRS delivered | 371 | 0.55 (0.16 – 1.87) | 0.34 |
| Time between latest cranial XRT and GK (per day) | 83 | 1.000 (0.998 – 1.002) | 0.78 |
| Prior delivery of extracranial XRT | 371 | 0.53 (0.25 – 1.12) | 0.10* |
| Time between latest extracranial XRT and GK (per day) | 78** | 0.99 (0.99 – 1.00) | 0.05* |
| Minimum time between any XRT to GK (per day) | 132** | 0.999 (0.997 – 1.001) | 0.18* |
| Time between dMRI and GK MRI (per day) | 371 | 1.01 (0.99 – 1.02) | 0.24 |
| dMRI Slice Thickness | 336 | 1.00 (0.82 – 1.21) | 0.98 |
| Number of metastases on dMRI (per metastasis) | 371 | 1.21 (1.06 – 1.38) | 0.0045* |
| Histology Breast vs Adenocarcinoma Melanoma vs Adenocarcinoma Neuroendocrine vs Adenocarcinoma Other vs Adenocarcinoma RCC vs Adenocarcinoma SCC vs Adenocarcinoma |
371 | 1.75 (0.78 – 3.95) 2.70 (1.30 – 5.61) 2.50 (0.70 – 8.86) 0.49 (0.11 – 2.27) 1.07 (0.36 – 3.18) 1.79 (0.63 – 5.11) |
0.08* |
| iBMV | 366 | 1.01 (1.00 – 1.02) | 0.15* |
*p < 0.20
**Removed from model selection procedure due to small sample size. Patients must have non-missing values for all variables to be included in the multivariable modeling.
Table 5B.
Multivariable analysis for factors significant for predicting both aBM and interval BM growth on GK MRI
| Multivariable Analysis | ||
|---|---|---|
| Characteristic | OR (95% CI) | p-value |
| Prior brain metastases resection | 3.94 (1.14 – 13.6) | 0.0301 |
| Number of metastases on dMRI (per metastasis) | 1.21 (1.06 – 1.38) | 0.0052 |
Impact of aBM and/or BM Interval Growth on SRS Management and Patient Outcomes
Significantly more BM were treated with GK (mean 2.8) than would have been predicted by dMRI (mean 2.1) (mean difference 0.7, 95% CI 0.54-0.83, p < 0.001). As such, patients with aBM were treated to a significantly larger mean target volume vs patients without aBM (4.2 vs 2.6 cm3, p = 0.001). Similarly, for patients with both aBM and BM growth, a significantly larger mean target volume was treated vs patients without either (5.1 cm3 vs 2.7 cm3, p < 0.001). Interval growth of known BM alone did not result in a difference in mean target volume (3.5 cm3 vs 2.8 cm3, p = 0.132)
For all patients, median follow-up from GK of 8.1 months (range 0-101.4 months) and a median imaging follow-up of 6.4 months from GK (range 0-116.4 months. Overall, median OS was 1.3 years and there was no significant difference in survival between patients with aBM (1.3 vs 1.3 years, p = 0.954) [Figure 1a)]. The discovery of aBM showed a trend for an increased incidence of distant intracranial failure when accounting for death as a competing risk [Figure 1b)], however, this was not statistically significant (p = 0.06). Similarly, the BM growth [Figure 1c)] or both aBM and BM growth [Figure 1e)] did not increase risk for distant intracranial failure and median OS was not significantly different in patients with BM growth vs those without [1.3 vs 1.6 years, p = 0.765, Figure 1d)], or with both aBM and BM growth [1.0 vs 1.4 years, p = 0.480, Figure 1f)]. The discovery of aBM led to change of management in only 1.8% of patients, most often for treatment with whole brain radiation (0.9%).
Figure 1.
a) Overall survival for patients with aBM vs those without. Median OS for each group was 1.3 years (p = 0.9540); b) Incidence of distant intracranial failure (DIF) for patients with aBM vs those without. The discovery of additional brain mets on GK MRI was not significantly associated with the incidence of distant intracranial failure when accounting for death as a competing risk (p-value = 0.0635); c) incidence of distant intracranial failure (DIF) for patients with BM growth vs those without. The discovery of BM growth on GK MRI was not significantly associated with the incidence of distant intracranial failure when accounting for death as a competing risk (p-value = 0.1302); d) overall survival for patients with BM growth vs those without. Median OS was not significantly different between those with BM growth vs those without: 1.3 vs 1.6 years, respectively (p = 0.7645); e) incidence of distant intracranial failure (DIF) for patients with both aBM and BM growth vs those without either. The discovery of BM growth on GK MRI was not significantly associated with the incidence of distant intracranial failure when accounting for death as a competing risk (p-value = 0.8069); f) overall survival for patients with both aBM and BM growth vs those without either. Median OS was not significantly different between those with BM growth vs those without: 1.0 vs 1.4 years, respectively (p = 0.4797).
DISCUSSION
SRS is the treatment of choice for a limited number of BM and offers feasibility and toxicity advantages over WBRT in both the up-front and salvage settings [7, 8, 10, 16, 25-27]. Recent prospective evidence has shown that up-front SRS is feasible for 5-10 lesions and leads to non-inferior outcomes when compared to patients with 2-4 lesions in terms of OS and toxicities [14]. Additionally, recent retrospective evidence has shown that patients treated with initial SRS for 5-15 lesions have similar survival, lower rates of salvage SRS, and no difference in rates of salvage WBRT vs those with 2-4 lesions [5]. Total lesion number, however, only appears to be part of the story and recently, cumulative lesion volume has been demonstrated to be at least as important a predictor of outcomes after SRS than total lesion number, if not more so [19-22].
Three studies have previously investigated factors predicting aBM found on GK MRI. In 2010, Nagai et al found that thin-slices and image acquisition delay after gadolinium contrast infusion resulted in greater lesion detection sensitivity for GK MRI over standard dMRI practices and therefore aBM. Since that time, Patel et al found that patients with multiple BM were at risk for having aBM on GK MRI and those with aBM had significantly worse median overall survival vs those without (12.1 vs 6.9 months) [18]. Most recently, Garcia et al found that the risk of finding aBM was associated with the number of BM found on dMRI and was inversely proportional to the size of the index brain metastasis [17]. Furthermore, in this study, finding aBM was unlikely to lead to treatment abortion, however, it did predict a higher incidence of distant intracranial failure but did not affect survival [17]. These studies, however, did not investigate the impact of metastasis growth in these patients in a larger sample of patients than the two more recent studies [17, 18]. We therefore sought to investigate factors associated with finding aBM on GK MRI and/or growth of previously identified lesions and correlate these findings with patient outcomes.
In our study, 32.9% of patients were found to have aBM on GK MRI, which is comparable to Nagai et al (33.7%) [28], but lower than both Garcia et al [17] and Patel et al [18] (49% and 41%, respectively). We also found a higher risk of aBM in patients who received prior systemic therapy on UVA, which contrasts the results seen in Garcia et al [17], which may serve as a surrogate for worse systemic control in these patients, however, this did not persist on MVA. On MVA, we found that aBM was significantly associated with iBMV. We also found that the number of metastases seen on dMRI is associated with an increased risk of finding aBM on GK MRI, which is consistent with findings from prior studies [17, 18, 29, 30]. Growth of previously identified BM was seen in 43.9% of our patients. iBMV was associated with growth of previously identified BMs and as expected, time between dMRI and GK MRI was significantly associated with interval growth. For each day between dMRI and GK MRI there was a 2% increase in risk of growth. On MVA, we also found that BM growth was more likely to be found in patients with the number of metastases found on initial dMRI (per metastasis). Both aBM and BM growth was found in 18.1% of patients. Patients with aBM or both aBM and BM growth had a significantly larger GK mean target volumes vs patients without. However, the discovery of aBM and/or BM growth led to change of management only 1.8% of the time, most often in favor of treatment with WBRT, a rate comparable to that seen in prior studies (~1.5%) [17]. In contrast to Garcia et al, however, we found that aBM did not predict worse intracranial failure but did trend towards significance (p = 0.06) [17]. Similarly, BM growth and both aBM plus BM growth did not show significantly worse intracranial failure vs patients without either and survival was not affected by either finding.
Our study has several limitations that we would like to acknowledge. Firstly, there are selection biases inherent to retrospective studies from a single academic center. Secondly, given that treatments took place at a large academic center, over more than a decade, several radiologists were responsible for noting growth of brain metastases and as such, inter-rater differences in criteria for responses are likely to exist. Thirdly, there was also variability in use/type of systemic therapy, intra/extracranial RT, histology, grade, other important factors between patients that may affect conclusions especially related to survival, though this heterogeneity reflects what is seen in other large academic centers and therefore the findings presented may be broadly applicable. Some of these factors, such as histologic subtypes or receptor patterns in the case of breast cancer, would likely cause even our current subgroups to show heterogeneity; initial attempts to fully subdivide these led to small patient numbers in any given group, and therefore we opted to keep broader groupings. Also, there are several technical aspects of MRI comparability that may obscure the assessment of changes in brain metastasis size, which include but are not limited to differences in slice thickness, magnet strength, and contrast volume/timing. A deeper analysis of these technical factors is planned for a future study. Lastly, intratumor hemorrhage was not considered as a distinct cause of BM growth largely because the degree to which a tumor appeared to grow from hemorrhage versus true progression could not be accurately determined.
In conclusion, aBM, BM growth, or both were found in 32.9%, 43.9%, and 18.1% of patient MRIs at time of GK, respectively, however, this only led to abortion of GK rarely. On average, one additional BM was found on GK MRI and finding aBM was more was predicted by higher number of metastases on dMRI and iBMV. Risk of interval BM growth was associated with increased time between MRIs, number of metastases on dMRI, and iBMV. Finding both aBM and BM growth was also associated with iBMV, when measured as a continuous variable, and more likely to be seen in patients with a greater time between dMRI and GK MRI, greater number of metastases on dMRI, longer time between latest extracranial XRT and GK, and those who previously underwent brain metastasectomy. Future studies from this dataset will focus on MRI technical factors, and a prospective validation of these findings controlling for variables that cannot be accurately controlled in a retrospective study. Identification of predictive factors for discovery of aBM and/or interval growth of brain metastases can provide valuable prognostic information, and aid in patient selection, counseling, or resource allocation, and highlight areas for quality improvement.
Supplementary Material
SUPPLEMENTAL MATERIALS
ACKNOWLEDGMENTS
Authors’ disclosure of potential conflict of interest
The authors have nothing to disclose.
Footnotes
Author contributions
Conception and design: Todd R. Mereniuk, Tim Lautenschlaeger, Gordon. A. Watson, Ryan M. Rhome
Data collection: Todd R. Mereniuk
Statistical analyses: Heather N. Burney
Data analysis and interpretation: Todd R. Mereniuk, Heather N. Burney, MS, Ryan M. Rhome
Manuscript writing: Todd R. Mereniuk, Ryan M. Rhome
Final approval of manuscript: Todd R. Mereniuk, Ryan M. Rhome
REFERENCES
- 1.Hamel-Perreault E, Mathieu D, Masson-Cote L. Factors influencing the outcome of stereotactic radiosurgery in patients with five or more brain metastases Curr Oncol 2019. Feb;26(1):e64-e69. doi: . Epub 2019 Feb 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Akeson P, Larsson EM, Kristoffersen DT, Jonsson E, Holtås S. Brain metastases--comparison of gadodiamide injection-enhanced MR imaging at standard and high dose, contrast-enhanced CT and non-contrast-enhanced MR imaging Acta Radiol 1995. May;36(3):300-6. [PubMed] [Google Scholar]
- 3.Sahgal A, Larson D, Knisely J. Stereotactic radiosurgery alone for brain metastases Lancet Oncol 2015. Mar;16(3):249-50. doi: (14) 71106-4. [DOI] [PubMed] [Google Scholar]
- 4.Tabouret E, Chinot O, Metellus P, Tallet A, Viens P, Gonçalves A. Recent trends in epidemiology of brain metastases: an overview Anticancer Res 2012. Nov;32(11):4655-62. [PubMed] [Google Scholar]
- 5.Hughes RT, Masters AH, McTyre ER, Farris MK, Chung C, Page BR, Kleinberg LR, Hepel J, Contessa JN, Chiang V, Ruiz J, Watabe K, Su J, Fiveash JB, Braunstein S, Chao S, Attia A, Ayala-Peacock DN, Chan MD. Initial SRS for Patients With 5 to 15 Brain Metastases: Results of a Multi-Institutional Experience Int J Radiat Oncol Biol Phys 2019. Aug 1;104(5):1091-1098. doi: . . [DOI] [PubMed] [Google Scholar]
- 6.Posner JB. Management of brain metastases Rev Neurol (Paris) 1992;148(6-7):477-87. [PubMed] [Google Scholar]
- 7.Sahgal A, Aoyama H, Kocher M, Neupane B, Collette S, Tago M, Shaw P, Beyene J, Chang EL. Phase 3 trials of stereotactic radiosurgery with or without whole-brain radiation therapy for 1 to 4 brain metastases: individual patient data meta-analysis Int J Radiat Oncol Biol Phys 2015. Mar 15;91(4):710-7. doi: . [DOI] [PubMed] [Google Scholar]
- 8.Sneed PK, Suh JH, Goetsch SJ, Sanghavi SN, Chappell R, Buatti JM, Regine WF, Weltman E, King VJ, Breneman JC, Sperduto PW, Mehta MP. A multi-institutional review of radiosurgery alone vs radiosurgery with whole brain radiotherapy as the initial management of brain metastases Int J Radiat Oncol Biol Phys 2002. Jul 1;53(3):519-26. doi: . [DOI] [PubMed] [Google Scholar]
- 9.Wen PY, Loeffler JS. Management of brain metastases Oncology (Williston Park) 1999. Jul;13(7):941-54, 957-61; discussion 961-2, 9. [PubMed] [Google Scholar]
- 10.Chang EL, Wefel JS, Hess KR, Allen PK, Lang FF, Kornguth DG, Arbuckle RB, Swint JM, Shiu AS, Maor MH, Meyers CA. Neurocognition in patients with brain metastases treated with radiosurgery or radiosurgery plus whole-brain irradiation: a randomised controlled trial Lancet Oncol 2009. Nov;10(11):1037-44. doi: . Epub 2009 Oct 2. [DOI] [PubMed] [Google Scholar]
- 11.Lukas RV, Gabikian P, Garza M, Chmura SJ. Treatment of brain metastases Oncology 2014;87(6):321-9. doi: . Epub 2014 Sep 6. [DOI] [PubMed] [Google Scholar]
- 12.Nieder C, Grosu AL, Gaspar LE. Stereotactic radiosurgery (SRS) for brain metastases: a systematic review Radiat Oncol 2014. Jul 12;9:155. . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yamamoto M, Serizawa T, Higuchi Y, Sato Y, Kawagishi J, Yamanaka K, Shuto T, Akabane A, Jokura H, Yomo S, Nagano O, Aoyama H. A Multi-institutional Prospective Observational Study of Stereotactic Radiosurgery for Patients With Multiple Brain Metastases (JLGK0901 Study Update): Irradiation-related Complications and Long-term Maintenance of Mini-Mental State Examination Scores Int J Radiat Oncol Biol Phys 2017. Sep 1;99(1):31-40. doi: . Epub 2017 Aug 7. [DOI] [PubMed] [Google Scholar]
- 14.Yamamoto M, Serizawa T, Shuto T, Akabane A, Higuchi Y, Kawagishi J, Yamanaka K, Sato Y, Jokura H, Yomo S, Nagano O, Kenai H, Moriki A, Suzuki S, Kida Y, Iwai Y, Hayashi M, Onishi H, Gondo M, Sato M, Akimitsu T, Kubo K, Kikuchi Y, Shibasaki T, Goto T, Takanashi M, Mori Y, Takakura K, Saeki N, Kunieda E, Aoyama H, Momoshima S, Tsuchiya K. Stereotactic radiosurgery for patients with multiple brain metastases (JLGK0901): a multi-institutional prospective observational study Lancet Oncol 2014. Apr;15(4):387-95. doi: . . [DOI] [PubMed] [Google Scholar]
- 15.Raman S, Chow R, Hoskin P, Chow E. How should radiation oncologists interpret the ASTRO evidence-based guideline and ASTRO Choosing Wisely campaign for the treatment of uncomplicated bone metastases? Pract Radiat Oncol 2017. Jan-Feb;7(1):13-15. doi: . Epub 2016 Sep 29. [DOI] [PubMed] [Google Scholar]
- 16.Shultz DB, Modlin LA, Jayachandran P, Von Eyben R, Gibbs IC, Choi CYH, Chang SD, Harsh GR, Li G, Adler JR, Hancock SL, Soltys SG. Repeat Courses of Stereotactic Radiosurgery (SRS), Deferring Whole-Brain Irradiation, for New Brain Metastases After Initial SRS Int J Radiat Oncol Biol Phys 2015. Aug 1;92(5):993-999. doi: . . [DOI] [PubMed] [Google Scholar]
- 17.Garcia MA, Lazar A, Duriseti S, Raleigh DR, Hess CP, Fogh SE, Barani IJ, Nakamura JL, Larson DA, Theodosopoulos P, McDermott M, Sneed PK, Braunstein S. Discovery of additional brain metastases on the day of stereotactic radiosurgery: risk factors and outcomes J Neurosurg 2017. Jun;126(6):1756-1763. doi: . Epub 2016 Jul 1. [DOI] [PubMed] [Google Scholar]
- 18.Patel TR, Ozturk AK, Knisely JP, Chiang VL. Implications of Identifying Additional Cerebral Metastases during Gamma Knife Radiosurgery Int J Surg Oncol 2012;2012:748284. doi: . . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hirshman BR, Wilson B, Ali MA, Proudfoot JA, Koiso T, Nagano O, Carter BS, Serizawa T, Yamamoto M, Chen CC. Superior Prognostic Value of Cumulative Intracranial Tumor Volume Relative to Largest Intracranial Tumor Volume for Stereotactic Radiosurgery-Treated Brain Metastasis Patients Neurosurgery 2018. Apr 1;82(4):473-480. doi: . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Baschnagel AM, Meyer KD, Chen PY, Krauss DJ, Olson RE, Pieper DR, Maitz AH, Ye H, Grills IS. Tumor volume as a predictor of survival and local control in patients with brain metastases treated with Gamma Knife surgery J Neurosurg 2013. Nov;119(5):1139-44. doi: . . [DOI] [PubMed] [Google Scholar]
- 21.Routman DM, Bian SX, Diao K, Liu JL, Yu C, Ye J, Zada G, Chang EL. The growing importance of lesion volume as a prognostic factor in patients with multiple brain metastases treated with stereotactic radiosurgery Cancer Med 2018. Mar;7(3):757-764. doi: . . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Soike MH, McTyre ER, Hughes RT, Farris M, Cramer CK, LeCompte MC, Lanier CM, Ruiz J, Su J, Watabe K, Bourland JD, Munley MT, O’Neill S, Laxton AW, Tatter SB, Chan MD. Initial brain metastasis velocity: does the rate at which cancers first seed the brain affect outcomes? J Neurooncol 2018. Sep;139(2):461-467. doi: . . [DOI] [PubMed] [Google Scholar]
- 23.McTyre ER, Soike MH, Farris M, Ayala-Peacock DN, Hepel JT, Page BR, Shen C, Kleinberg L, Contessa JN, Corso C, Chiang V, Henson-Masters A, Cramer CK, Ruiz J, Pasche B, Watabe K, D’Agostino R, Su J, Laxton AW, Tatter SB, Fiveash JB, Ahluwalia M, Kotecha R, Chao ST, Braunstein SE, Attia A, Chung C, Chan MD. Multi-institutional validation of brain metastasis velocity, a recently defined predictor of outcomes following stereotactic radiosurgery Radiother Oncol 2020. Jan;142:168-174. doi: . . [DOI] [PubMed] [Google Scholar]
- 24.Aoyama H, Shirato H, Tago M, Nakagawa K, Toyoda T, Hatano K, Kenjyo M, Oya N, Hirota S, Shioura H, Kunieda E, Inomata T, Hayakawa K, Katoh N, Kobashi G. Stereotactic radiosurgery plus whole-brain radiation therapy vs stereotactic radiosurgery alone for treatment of brain metastases: a randomized controlled trial JAMA 2006. Jun 7;295(21):2483-91. doi: . [DOI] [PubMed] [Google Scholar]
- 25.Aoyama H, Tago M, Kato N, Toyoda T, Kenjyo M, Hirota S, Shioura H, Inomata T, Kunieda E, Hayakawa K, Nakagawa K, Kobashi G, Shirato H. Neurocognitive function of patients with brain metastasis who received either whole brain radiotherapy plus stereotactic radiosurgery or radiosurgery alone Int J Radiat Oncol Biol Phys 2007. Aug 1;68(5):1388-95. doi: . [DOI] [PubMed] [Google Scholar]
- 26.Brown PD, Jaeckle K, Ballman KV, Farace E, Cerhan JH, Anderson SK, Carrero XW, Barker FG, Deming R, Burri SH, Ménard C, Chung C, Stieber VW, Pollock BE, Galanis E, Buckner JC, Asher AL. Effect of Radiosurgery Alone vs Radiosurgery With Whole Brain Radiation Therapy on Cognitive Function in Patients With 1 to 3 Brain Metastases: A Randomized Clinical Trial JAMA 2016. Jul 26;316(4):401-409. doi: . Erratum in: JAMA 2018. Aug 7;320(5): 510. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Nagai A, Shibamoto Y, Mori Y, Hashizume C, Hagiwara M, Kobayashi T. Increases in the number of brain metastases detected at frame-fixed, thin-slice MRI for gamma knife surgery planning Neuro Oncol 2010. Nov;12(11):1187-92. doi: . . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Donahue BR, Goldberg JD, Golfinos JG, Knopp EA, Comiskey J, Rush SC, Han K, Mukhi V, Cooper JS. Importance of MR technique for stereotactic radiosurgery Neuro Oncol 2003. Oct;5(4):268-74. doi: . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Engh JA, Flickinger JC, Niranjan A, Amin DV, Kondziolka DS, Lunsford LD. Optimizing intracranial metastasis detection for stereotactic radiosurgery Stereotact Funct Neurosurg 2007;85(4):162-8. doi: . . [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
SUPPLEMENTAL MATERIALS