Optimal timing of radiotherapy following brain metastases surgery

Abstract

Background

There is growing evidence supporting the need for a short time delay before starting radiotherapy (RT) treatment postsurgery for most optimal responses. The timing of RT initiation and effects on outcomes have been evaluated in a variety of malignancies, but the relationship remains to be well established for brain metastasis.

Methods

Retrospective study of 176 patients (aged 18–89 years) with brain metastases at a single institution (March 2009 to August 2018) who received RT following surgical resection. Time interval (≤22 and >22 days) from surgical resection to initiation of RT and any potential impact on patient outcomes were assessed.

Results

Patients who underwent RT >22 days after surgical resection had a decreased risk for all-cause mortality of 47.2% (95% CI: 8.60, 69.5%). Additionally, waiting >40 days for RT after surgical resection more than doubled the risk of tumor progression; adjusted hazard ratio 2.02 (95% CI: 1.12, 3.64).

Conclusions

Findings indicate that a short interval delay (>22 days) following surgical resection is required before RT initiation for optimal treatment effects in brain metastasis. Our timing of RT postsurgical resection data adds definition to current heterogeneity in RT timing, which is especially important for standardized clinical trial design and patient outcomes.

Brain metastases have been estimated to occur in 10% of all cancer diagnoses, and are increasingly more common as new therapeutics, improved imaging, and screening techniques have led to longer survival after a primary cancer diagnosis.¹ Incidence, number, location, and survival of metastases vary greatly based on the primary tumor.¹ Collectively, primary tumor and resultant metastases cause significant morbidity and mortality.² Evidence suggests that the optimal treatment of brain metastases consists of surgical resection and adjuvant radiotherapy (RT); a method that has been increasingly used as an alternative to whole-brain radiotherapy (WBRT).^3–5 WBRT improves tumor control at the resection cavity and at distant brain sites without extending overall survival (OS) or functional independence.⁴ An alternative to WBRT is stereotactic radiosurgery (SRS), which utilizes highly precise delivery of radiation to specific targets, and therefore offers the benefit of reduced harm to neurocognition and patient quality of life.³ SRS also offers the advantage of single-session delivery, which is important considering the timing delays of chemotherapy in patients with newly diagnosed or progressive systemic diseases.³ Though this targeted therapy has increased OS, the optimal timing of radiation postresection is not well characterized and needs to be better understood in efforts to increase patient survival.

While timing of RT and effects on outcomes have been evaluated in a variety of malignancies, including glioblastoma,⁶ the relationship has not been well established for brain metastasis. Further, postresection tumor cavity dynamics are still being elucidated. A rationale to delay SRS for up to 6 weeks following surgical resection was to allow time for patient recovery from surgery, cavity shrinkage to reduce treatment volume, and data that few metastases recur after 6 weeks if gross-total resection has been achieved.⁷ A retrospective study revealed that any delay >3 weeks between resection and SRS was significantly associated with higher rates of treatment failure, with estimated 6-month control rates dropping from 94% to 77%.⁷ This longer surgery-to-SRS delay was also inversely proportional to resection cavity size and directly proportional to margin dose.⁷ Another study looking at the effect of timing of concurrent chemoradiation in patients with glioblastoma revealed that a modest delay of 4 weeks is associated with improved survival outcomes.⁸ This was further validated for glioblastoma in a study that revealed chemoradiation therapy delays >5 weeks in patients with glioblastoma did not negatively affect outcomes, while initiation before 3 weeks was shown to be detrimental.⁶ Here, we retrospectively explore the impact of optimal RT timing and the role of RT treatment following surgical resection of brain metastases. We aimed to assess timing intervals to determine any differences in OS and progression-free survival (PFS).

Methods

Study Design and Setting

This is a retrospective, noninterventional medical chart review study to evaluate optimal RT timing and role of RT treatment following surgical resection of brain metastases, at a single institution. Patients between the ages of 18 and 89 years with brain metastases who received RT treatment postoperatively between March 2009 and August 2018 were included. Patients whose precise timing of RT treatment initiation was unable to be determined were excluded. This study was approved by the Institutional Review Board.

Indications for Brain Metastasis Resection

At our institution we followed guidelines for good practice in the treatment of brain metastasis. Careful selection is employed as it is well reported in the literature not all brain metastasis patients are likely to benefit from surgery.⁹ Patients were selected for surgery on a case-by-case basis. The majority of patients indicated for surgery in this study had 1–3 brain lesions, with favorable surgical accessibility (only 4 located at the skull base, and 4 located in deep brain structures; basal ganglia and pons). Patients had at least 1 lesion with local symptomatic mass effect with good localization on neurologic exam, or undiagnosed primary site of cancer. The majority of patients indicated for surgery had a Karnofsky performance status (KPS) >70; however, patients with a lower KPS were not automictically excluded, and control of extracranial disease was also considered. Patients indicated for surgery had absence of leptomeningeal involvement at the time of surgical resection.

SRS Planning Techniques

Patients were immobilized using headrests and customized thermoplastic face masks. Computed tomography (CT) scans of each patient’s cranial anatomy were acquired using 1-mm slices. These were fused with specific magnetic resonance imaging (MRI) T1-weighted postgadolinium axial series that were obtained with 1-mm slice thickness. Eclipse planning software (Varian Medical Systems, Palo Alto, CA) was used to fuse MRI series to CT simulation scans. Brain MRI facilitated delineation of the clinical target volumes (CTVs) to outline areas of enhancement that correlated to the brain metastasis resection cavities. Planning target volumes (PTVs) were created by expanding the CTV structures by 2 mm. If preoperatively the tumor was in contact with dura, these areas included a total expansion of 5 mm PTV. Normal structures were delineated to use as avoidance structures during planning to minimize radiation-associated toxicity. Treatment plans were created using either iPlan (RT Image, Brainlab, Munich, Germany) or Eclipse planning software (Varian Medical Systems, Palo Alto, CA). Cavity size dictated the use of 3–5 fractions for radiation treatment with larger cavities utilizing more fractionated courses. Cavities with volumes >30 cc were unlikely to meet normal brain tissue dose constraints for 3-fraction regimens (V24 Gy <16.8 cc), and thus required transition to use of a 5-fraction plan.¹⁰ Cavities near eloquent locations including the brainstem, motor strip, or optic nerves were also more likely to receive treatment with 5 fractions to decrease risk of radiation-induced toxicity. Additionally, the use of 3- versus 5-fraction regimens is currently being assessed in the ALLIANCE A071801 trial. All patients were treated on a linear accelerator (LINAC) with 6 MV photons and a multileaf collimator (MLC) with 2.5 or 5 mm leaves. Plans utilized 4–5 noncoplanar arcs with the goal for a hot spot of at least 120% of the prescription dose. Each plan was checked by physics staff for quality assurance prior to the patient undergoing treatment. Patient positioning for each day of treatment was verified using image guidance with cone-beam CT along with ExacTrac (Brain Lab, Munich, Germany) optical surface monitoring for detection of intrafraction motion detection.

Measurements and Outcomes

A medical chart review was conducted to collect patient demographics, KPS, treatment characteristics, cranial imaging findings, and clinical outcomes. RT characteristics collected included radiation dose in Gray (Gy) and number of fractions received. Lesion treatment response was determined by examining contrast enhancement volume and radiographic presence of the treated lesions on follow-up imaging. Surgical resection status (partial vs total), number of resected and nonresected lesions, and final histology were also documented. Local progression was defined at the outset of the study as a persistent increase in the volume of contrast enhancement on MRI, histological confirmation of disease recurrence, or lesions requiring treatment. To add further granularity to this definition, local progression was determined by radiographic progression of disease within or subjacent to the resection cavity. Disease progression include both new or increased contrast enhancement, along with increased size of the lesion. The determination of recurrence was made by senior radiologist in the final radiology report. All relevant radiology reports were assessed (both in findings and impressions) for language supporting disease progression within or subjacent to resection cavity. If the radiology report documented suspicion for progression, but not definitive for progression, the next follow-up MRI was reviewed, and if definitive progression identified, the prior imaging date for earliest sign of progression was collected as the date of local progression. In cases of multiple resections (on the same operative date), local progression was determined by first radiographic sign of progression at any site. Data regarding concurrent treatment, as well as extracranial disease status were also assessed, as was the cause of death when documented. Categorization of continuous variables was performed using methodology consistent with previous studies.

Primary outcome of interest was OS, and secondary outcome of interest was PFS. Intervals from the time of surgical resection to RT initiation were compared separately for each of these outcomes. Exploratory analysis for OS began with plotting the months each patient survived postoperatively versus the time from resection to RT initiation. A recursive partitioning analysis was then performed using deletion, substitution, and addition for the interval from resection to initiation of RT using the Brier method to incorporate the median OS for the entire study population, 300 days. A multivariable Cox analysis incorporating stabilized inverse probability weights (IPWs) was used to discern variables predictive of OS. Scatterplot analysis was used to assess the relationship between time to RT initiation and PFS in months. A recursive partitioning analysis was performed using deletion, substitution, and addition for the interval from resection to initiation of RT using the Brier method to incorporate the median PFS for the entire study population.

Statistical Analysis

Dichotomous and categorical variables were reported as counts and percentages, and compared between groups using Fisher’s exact test. Continuous variables were assessed for normality of distribution via the Shapiro–Wilk test. Accordingly, all continuous variables are described using medians and interquartile ranges and compared between groups using Kruskal–Wallis tests. No observations contained missing values.

For each outcome of interest, a right-censored Kaplan–Meier (KM) curve was plotted using time-to-event data, with participants censored at the time of last documented clinical follow-up if the outcome of interest had not occurred. Recursive partitioning analysis via a deletion, substitution, and addition algorithm was then performed to determine the cut point for time to radiation therapy that was most statistically significantly associated with each outcome using Brier scoring.¹¹ This methodology permits all events to be included in the partitioning algorithm, with observations censored after the median study time included in the estimation of the partitioning parameter in a weighted manner. The resulting KM curves were then stratified by this interval and compared using the log-rank test.

To account for observed confounding and to address the sample size imbalance between exposure groups, IPWs were used. Specifically, each patient’s age, sex, RT dose per 5 Gy, concurrent treatment regimen, KPS, and extracranial disease status were used to calculate propensity for early versus delayed RT initiation from a nonparsimonious logistic regression model. Propensity score calculations for PFS also included histology and surgical resection status as covariates. Balance of included covariates and region of common support were then assessed using standardized mean differences, jitter plots, and histograms. A stabilized inverse probability-weighted multivariable Cox proportional hazards regression model was then constructed from candidate covariates which were known or suspected risk factors for each outcome, provided they were not included in the propensity model. Least Absolute Shrinkage and Selection Operator (LASSO) regression was employed to identify those factors most associated with the outcomes of interest. Briefly, LASSO regression adds a penalty term which shrinks the coefficient estimates for factors weakly associated with the outcome toward 0, resulting in a parsimonious model of only covariates strongly associated with the outcome. Assumptions required for proportional hazards regression were then assessed numerically for each covariate, and graphically by the plotting of scaled Schoenfeld residuals. An interaction term was created between each covariate and time for the overall hazard model, ensuring each covariate was not time-dependent. The goodness of fit of each model was then assessed using Cox–Snell residuals. Crude (univariable) and adjusted (multivariable) inverse probability-weighted results are presented as a hazard ratio (HR) and 95% confidence interval (CI) calculated using robust standard error.

A P value <.05 was considered to be statistically significant. All tests performed were 2-sided. All analyses were performed using R software, version 4.1.0 (R Foundation for Statistical Computing, Vienna, Austria). The final analysis was confirmed using Stata, version 17.0 (StataCorp, College Station, TX).

Results

The final analytic sample included 176 patients; 16 patients were excluded from the PFS analysis as their clinical disease was not evaluated past treatment staging MRIs. Final PFS analytic sample size was 160 patients.

Data analysis indicated that the OS days for patients whose radiation was initiated ≤22 days after resection were statistically significantly different from those whose radiation was initiated at >22 days (Supplementary Figure 1 and Table 1). The difference in OS days between ≤22 days 190.50 days [101.00, 321.00] versus >22 days 338.00 days [183.50, 573.00], P = .004 was significant despite overall comparable patient demographics and clinical characteristics. The only notable difference between groups was that patients who received RT at ≤22 days postoperatively were more frequently treated with immunotherapy (26.7%), compared to patients whose RT was >22 days (7.5%) (Table 1). For this reason, IPWs generated for OS incorporated concurrent therapy as a covariate and significant balance was attained.

Table 1.

Baseline Patient Demographics and Clinical Characteristics Stratified by the Number of Days Between Surgical Resection and Initiation of Radiotherapy (RT)

Variable	Level	Total	Time to RT ≤ 22 Days	Time to RT > 22 Days	P Value
n		176	30	146
Age at surgery (years)		62.00 [53.75, 72.00]	62.00 [53.25, 73.00]	62.00 [54.00, 71.75]	.725
Sex	Male	96 (54.5)	17 (56.7)	79 (54.1)	.843
	Female	80 (45.5)	13 (43.3)	67 (45.9)
Time to RT (days)		35.00 [27.00, 44.00]	19.00 [17.00, 21.00]	38.00 [30.25, 46.75]	<.001
RT dose (Gy)		30.00 [27.00, 30.00]	30.00 [25.50, 30.00]	30.00 [27.00, 30.00]	.338
RT fractions		5.00 [5.00, 5.00]	5.00 [4.25, 5.00]	5.00 [5.00, 5.00]	.866
Categorical RT fractions	≤3 fractions	39 (22.2)	6 (20.0)	33 (22.6)	1.000
	>3 fractions	137 (77.8)	24 (80.0)	113 (77.4)
Number of resected lesions	1 Lesion	148 (84.1)	27 (90.0)	121 (82.9)	.083
	2 Lesions	23 (13.1)	1 (3.3)	22 (15.1)
	3 Lesions	5 (2.8)	2 (6.7)	3 (2.1)
Number of nonresected lesions	None	111 (63.1)	15 (50.0)	96 (65.8)	.145
	≥1 Lesion	65 (36.9)	15 (50.0)	50 (34.2)
Extracranial disease	Stable	133 (75.6)	22 (73.3)	111 (76.0)	.816
	Progressive	43 (24.4)	8 (26.7)	35 (24.0)
Concurrent treatment	Chemotherapy	23 (13.1)	4 (13.3)	19 (13.0)	.023
	Immunotherapy	19 (10.8)	8 (26.7)	11 (7.5)
	Neither	133 (75.6)	18 (60.0)	115 (78.8)
	Both	1 (0.6)	0 (0.0)	1 (0.7)
Surgical resection	Partial	16 (9.1)	2 (6.7)	14 (9.6)	1.000
	Total	160 (90.9)	28 (93.3)	132 (90.4)
Histology	Lung	66 (37.5)	9 (30.0)	57 (39.0)	.719
	Breast	12 (6.8)	3 (10.0)	9 (6.2)
	Colon	9 (5.1)	1 (3.3)	8 (5.5)
	Melanoma	31 (17.6)	6 (20.0)	25 (17.1)
	Ovarian	6 (3.4)	2 (6.7)	4 (2.7)
	Renal	12 (6.8)	1 (3.3)	11 (7.5)
	Other	40 (22.7)	8 (26.7)	32 (21.9)
KPS	40	4 (2.3)	0 (0.0)	4 (2.7)	.175
	50	10 (5.7)	3 (10.0)	7 (4.8)
	60	6 (3.4)	2 (6.7)	4 (2.7)
	70	27 (15.3)	8 (26.7)	19 (13.0)
	80	48 (27.3)	6 (20.0)	42 (28.8)
	90	72 (40.9)	9 (30.0)	63 (43.2)
	100	9 (5.1)	2 (6.7)	7 (4.8)
Outcome	Alive	81 (46.0)	10 (33.3)	71 (48.6)	.160
	Deceased	95 (54.0)	20 (66.7)	75 (51.4)
Overall survival (days)		299.50 [170.00, 540.00]	190.50 [101.00, 321.00]	338.00 [183.50, 573.00]	.004
Progression	Disease-free	121 (68.8)	24 (80.0)	97 (66.4)	.195
	Local progression	55 (31.2)	6 (20.0)	49 (33.6)
Progression-free survival (days)		210.50 [118.75, 363.25]	137.00 [62.75, 251.25]	232.00 [127.00, 372.75]	.018

Abbreviations: Gy, Gray; KPS, Karnofsky performance status; N/A, not applicable.

KM plots for OS stratified by interval to RT initiation relative to the cut point of 22 days confirmed this overall difference in survival (P = .0061; Figure 1). The number of RT fractions was found to be a statistically significant predictor in the LASSO regression models, but was found to violate the assumptions of proportional hazards and was thus excluded from the final model. After adjusting for the number of nonresected lesions and tumor histology, the HR for all-cause mortality for those whose RT was initiated ≤22 days was 1.89 (95% CI: 1.09, 3.28) times the risk of mortality for those whose RT was initiated at >22 days (Table 2; Figure 2). This indicates that delaying RT initiation by at least 22 days decreases all-cause mortality by 47.2% (95% CI: 8.60, 69.5%).

Figure 1.

Kaplan–Meier curve for overall survival including all patients, stratified by the time between surgical resection and initiation of radiation therapy.

Table 2.

Stabilized Inverse Probability-Weighted Multivariable Cox Proportional Hazards Regression Analysis for Clinical Characteristics Associated With Overall Survival

		Univariablea			Multivariablea
Variable	Level	HR	95% CI	P Value	HR	95% CI	P Value
Time to RT (days)	>22 days	Reference	Reference	Reference	Reference	Reference	Reference
	≤22 days	1.86	1.08–3.20	.024	1.89	1.09–3.28	.023
Number of nonresected lesions	None	Ref	Ref	Ref	Ref	Ref	Ref
	≥1 Lesion	1.59	1.05–2.41	.028	1.66	1.08–2.56	.021
Histology	Lung	Ref	Ref	Ref	Ref	Ref	Ref
	Breast	0.57	0.18–1.75	.323	0.63	0.21–1.90	.416
	Colon	0.89	0.26–3.01	.855	1.18	0.35–4.02	.788
	Melanoma	1.01	0.55–1.84	.976	0.95	0.53–1.73	.875
	Ovarian	1.07	0.54–2.11	.848	0.84	0.38–1.89	.680
	Renal	0.52	0.23–1.20	.125	0.55	0.22–1.39	.203
	Other	1.48	0.89–2.47	.132	1.57	0.93–2.65	.090

Abbreviations: CI, confidence interval; HR, hazard ratio; KPS, Karnofsky performance status; RT, radiotherapy.

^aStabilized inverse probability weights account for age, sex, RT dose per 5 Gy, concurrent treatment regimen, KPS, and extracranial disease status.

Figure 2.

Forest plot demonstrating stabilized inverse probability-weighted hazard ratios from a multivariable analysis of clinical characteristics associated with overall survival.

Scatterplot analysis and recursive partitioning analysis for the entire study population, 211 days (Supplementary Figure 2), indicated that patient PFS, for RT initiated ≤40 days was statistically significantly different from those whose RT was initiated >40 days; despite overall comparable patient demographics and clinical characteristics.

KM plots for PFS validated this PFS difference (P = .0025) when dichotomizing using 40 days as the dividing point (Figure 3). Multivariable Cox analysis using updated stabilized IPWs to determine factors associated with PFS showed that only the dichotomized interval variable was a significant covariate—all other statistically significant predictors were accounted for in the observation weights. Therefore, the adjusted HR for progression in patients whose RT is initiated >40 days following surgical resection is 2.02 (95% CI: 1.12, 3.64) times the risk for progression in patients whose RT is initiated ≤40 days postoperatively (Supplementary Table 1; Figure 4).

Figure 3.

Kaplan–Meier curve for progression-free survival, stratified by the time between surgical resection and initiation of radiation therapy (n = 160).

Figure 4.

Forest plot demonstrating stabilized inverse probability-weighted hazard ratios from a multivariable analysis of clinical characteristics associated with progression-free survival (n = 160).

Discussion

In patients with systemic cancer with 1 or more brain metastases, treatment modalities include surgical resection, WBRT, radiosurgery, and chemotherapy. As treatment options increase and complexity grows, gaps in evidence needed to make treatment decisions regarding the timing of adjuvant therapies following surgical resection have also developed. Few studies have investigated optimal timing of RT. However, some studies have revealed important information regarding the variance between treatment efficacies dependent on postresection timing.^12–20

The idea of optimal RT timing suggests an optimal state of postsurgical cavity, one that has been shown in studies to fluctuate in size and characteristics.¹² There are consequences associated with this timing, such as contraction of the surgical cavity. Increased time from surgery allows for more contraction, which leads to a decrease in the volume of normal tissue being irradiated, and therefore reduced risk of symptomatic radiation necrosis. Though, this delay in radiation and systemic treatment may translate into recurrent or progressive disease, a concern that is particularly important in rapidly progressive malignancies. One clinical trial revealed that the greatest volume change occurred immediately after surgery in postoperative days 0–3 with no statistically significant volume change occurring up to 33 days after surgery in most patients.¹² This suggests no benefit of cavity shrinkage in waiting >3 days to perform cavity SRS.

Another study observed a significant volume reduction for larger tumors during the intermediate phase of 22–42 days postoperatively.¹³ The authors cautioned not to treat any cavities in intervals earlier than 21 days after surgery due to the risk of irradiating more normal tissue.¹³

Another retrospective study found that after resection, despite there being a decrease in target volume due to surgical intervention, there is overall cavity growth.¹⁵ The authors reported that small cavities experience the greatest percentage growth versus large postoperative cavities, which experience minimal cavity change. These differences might be related to patient or tumor factors, and one of these studies found that younger patients and tumors with dural involvement have a higher likelihood of cavity shrinkage.¹³

One hypothesis to explain these variances in radiation efficacy suggests the tissue biomechanisms responsible involve hypoxia and edema of the surgical cavity. These local changes in surrounding brain tissue can diminish radiosensitivity of the region, thus making treatment ineffective without consideration of the pathological time line. While the time line of postoperative ischemia is not well understood, studies have shown that it is also a driver of infiltrative tumor growth in glioblastoma patients,^16,17 suggesting there might be an underlying role of ischemia in tumor control and OS.

A study comparing preoperative versus postoperative stereotactic body radiation therapy (SBRT) found that distal site brain tumor growth was predominant over local growth in postoperative SBRT patients.¹⁸ The authors also reported that leptomeningeal dissemination was the most common pattern of treatment failure with this risk being higher when compared to postoperative WBRT or definitive SRS. It is possible that this occurs intraoperatively via cerebrospinal fluid, though it has been proposed that preoperative SRS may lower this risk by sterilizing the resection area.¹⁴

Another study recently published by O’Brien et al. looked at local recurrence after adjuvant SRS in what they describe as “real-world” conditions.¹⁹ This group found a significant number of patients that were referred for SRS were treated >8 weeks postoperatively or not treated at all. In the group of patients who were treated >8 weeks postoperatively, their local recurrence risk was equivalent to that of patients who never received SRS. Additionally, another recent retrospective study by Bander et al. examined adjuvant SRS timing in postmetastasectomy patients and found that SRS delivered within 1 month had a lower surgical site recurrence rate (6.1% at 1 year) compared to SRS delivered 1–2 months postsurgery (9.2%) or if delivered >2 months postsurgery (27.3%).²⁰ Similarly, OS in their patient cohort was significantly lower if radiation was delayed over 2 months.

While multiple recent studies comment on delays in RT initiation postmetastasis resection leading to worse OS and higher rates of progression, few comment on when would be too soon to deliver SRS to the brain metastasis resection cavity.

Here we report an optimal timing for initiation of brain SRS > 22 days after surgical resection for initiation of RT. We show that delaying RT by at least 22 days decreases overall mortality by 47.2%. Taking into account previous reports and the data we present here, this could be interpreted as a narrow 23–30 days window for optimal postoperative brain RT delivery in the treatment of metastasis. This could present some patient care coordination challenges; such a treatment window may be difficult to achieve for all patients. However, adoption of standardization of RT timing to a 23- to 30-day window postresection of brain metastasis could help minimize current RT timing heterogeneity, and will be especially important for future clinical trial design and subsequent improved patient outcomes.

Limitations

The study is limited by factors inherent to a retrospective study, including related patient selection. Additionally included in this study are all various pathologies of brain metastasis, and OS can be highly influenced by aggressiveness of the primary tumor pathology. Likewise, time to radiation is affected by numerous social and clinical variables, some of which we attempted to control for in our multivariable analysis. Regarding the only statistically significant difference between our groups, we have 2 hypotheses regarding why patients who received RT at ≤22 days postoperatively were more frequently treated with immunotherapy (26.7%), compared to patients whose RT was >22 days (7.5%). First, many of the patients included in this study who received immunotherapy had the diagnosis of metastatic melanoma or non-small cell lung carcinoma and participated in clinical trials to receive immunotherapy. Therefore, we believe the patients that were eligible for a clinical trial and wanted to participate in the trial had expedited care by nature of requirements to follow study protocols. Being enrolled in a clinical trial may have directly contributed to closer involvement of the treating physicians and more rapid follow-up to begin therapies, including RT. The alternative though related hypothesis for this observed association is socioeconomic in nature. Patients who had greater access to medical care, whether through physical location living near to a tertiary medical center, or having healthcare insurance which qualifies a patient for access to immunotherapeutic treatments, potentially results in receiving more rapid treatments including RT through access and treatment availability. We acknowledge the potential confounding nature of this association, and for this reason have statistically adjusted for these differences in sample comparisons to allow our reported finding to account for such differences between the groups.

Conclusions

It was previously unclear for brain metastasis if a short delay in RT initiation postsurgery is needed, as previously demonstrated for glioblastoma. Recent studies suggest waiting > 1 month to initiate RT for brain metastasis is overall detrimental to patients with decreased OS and PFS; this is supported by the data we present here. Further, we posit that an optimal RT treatment window could be 23–30 days. We suggest delaying RT initiation to treat brain metastasis by at least 22 days after surgical resection to minimize radiation complications and surrounding tissue radiation exposure.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of interest statement. The authors declare that they have no conflicts of interest regarding the current work.