Leukoencephalopathy in patients with brain metastases who received radiosurgery with or without whole brain radiotherapy

Whole brain radiation therapy (WBRT) for brain metastases (BMs) is a common cause of radiation-induced leukoencephalopathy; however the safety of alternative stereotactic radiosurgery (SRS) remains unclear. This study examined the incidence of leukoencephalopathy in patients treated with SRS alone versus WBRT plus SRS for BMs with a focus on the relationship between prognostic factors and leukoencephalopathy. Analysis was performed between 2002 and 2021. The total enrollment was 993 patients with the distribution: WBRT plus SRS (n = 291) and SRS only (n = 702). Leukoencephalopathy was graded from 0 to 3 for changes in white matter indicated by the MRI after WBRT or SRS. Patient characteristics and SRS dosimetric parameters were reviewed to identify factors that contributed to the incidence of leukoencephalopathy or overall survival. The incidence of leukoencephalopathy was consistently higher in WBRT plus SRS group than in SRS alone group (p < 0.001). Leukoencephalopathy was also associated with a larger total tumor volume (≧28cm3; p = 0.028) and age (> 77 years; p = 0.025). Nonetheless, the SRS integral dose to skull in the subgroup of WBRT plus SRS treatment was not demonstrated significance in development of leukoencephalopathy (p = 0.986 for integral dose 1–2 J, p = 0.776 for integral dose > 2 J). This study revealed that SRS is safe for oligo-BMs in terms of leukoencephalopathy development. Patient age and total tumor volume were identified as important factors in assessing the development of leukoencephalopathy. The additional of SRS (even at an integral dose > 2 J) did not increase the incidence of leukoencephalopathy.


Introduction
Approximately 30% of patients with cancer develop brain metastases (BMs) [1,2]. Whole brain radiation therapy (WBRT) has long been the standard treatment for intracranial metastases [3,4]; however, this has raised concerns pertaining to physical side effects (changes in white matter; leukoencephalopathy) and radiation-induced neurocognitive dysfunction [5,6]. This has led to the development of an alternative treatment involving stereotactic radiosurgery (SRS).
The safety of SRS in treating BMs has been assessed in two retrospective studies. Those works revealed that the risk of developing progressive radiation-induced leukoencephalopathy was lower for patients who underwent SRS than among those who underwent WBRT in conjunction with SRS [7,8]. They identified a number of risk factors for radiation-induced leukoencephalopathy, including total tumor number, total tumor volume, and integral doses; however, there were a number of inconsistencies between the papers. Our objective in the current study was to compare the incidence of leukoencephalopathy between patients treated using SRS alone or WBRT plus SRS for BMs. We also sought to identify the dosimetric parameters and prognostic factors associated with leukoencephalopathy.

Patient population
This retrospective study included all patients (n = 933) who received SRS or WBRT as the initial treatment for BMs between 2002 and 2021. Among those patients, 291 patients received WBRT plus SRS whereas 702 received SRS alone. Eligibility criteria included newly diagnosed BMs at the time of the first SRS or WBRT session, pathologic confirmation of extracerebral tumor site (e.g., lung or breast) from either the primary site or a metastatic lesion, and no contraindications for magnetic resonance imaging (MRI). Exclusion criteria included previous treatment for nonmetastatic brain disease, any contraindication to MRI, unavailable follow-up clinical records or imaging studies, or a history of a demyelinating disorder such as multiple sclerosis. The institutional review board approved the current study prior to initiation and written informed consent was obtained from each patient.

SRS treatment
SRS treatment was performed using a Leksell Gamma Knife (model 3C or 4C) with a 60 Co sources in conjunction with a Perfexion radiosurgery unit (Elekta Instruments) in a single session. Radiosurgery plans were created using Leksell Gamma Plan software and reviewed by a neurosurgeon, radiation oncologist, and medical physicist prior to radiosurgery. The median margin dose was 19 Gy (range 11-30 Gy) and the median radiation volume was 2.29 cm 3 (range 0.01-57.94 cm 3 ). The isodose line around the tumor was between 45 and 90% of the maximum dose. Patients who underwent WBRT received a median margin dose of 30 Gy (range 16.7-44.4 Gy) administered over a median of 14 fractions [9]. None of the patients in this series underwent a second course of WBRT.

Clinical presentation and neuroimaging assessments
Prior to registration, each patient was assessed in terms of medical history, physical condition, and neurological condition. MRIs were performed at the time of SRS or WBRT and at least once as a follow-up. Clinical data were obtained for all eligible patients, including age, sex, histology and primary site of tumors, extracranial disease, and Karnofsky performance status (KPS) as well as previous treatments, including systemic anti-cancer medication or craniotomy. After treatment, SRS-related toxicity, WBRT dose and fractionation, tumor volume, tumor location, SRS dosage, integral dose to the brain, and survival were recorded. The dates of diagnosis, radiosurgical treatment, previous or additional leukoencephalopathy, and the last follow-up visit were included in the analyses. The observation period was divided into the following three intervals: before radiosurgery, the interval from diagnosis to the first time radiosurgery; the latency period, the interval from the first time radiosurgery to leukoencephalopathy initiation; and after leukoencephalopathy onset, the interval from leukoencephalopathy initiation to the end of the follow-up period. The primary end point was last clinical follow-up visit. Secondary end points was time to death, but the cause of death and cause-specific survival was not assessed.
All patients underwent MRI follow-ups at 3-month intervals. All MRI scans (T1-weighted [T1W], T2-weighted [T2W], and contrast-enhanced T1W images) were reviewed by the same group of investigators with a focus on changes in cerebral white matter, local tumor control, distant tumor control, and radiation necrosis. In some situations, it is preferable to use FLAIR sequences to identify changes in periventricular white matter, as the sensitivity is higher than that of T2 sequences [10]. We employed a simplified grading system, which has been previously described [11] and clinically validated [12][13][14], for the detection of changes in cerebral white matter. Under this grading system, T2 sequences are categorized as follows: grade 0 (normal), grade 1 (mild changes in periventricular white matter), grade 2 (moderate changes in periventricular white matter), and grade 3 (severe and/or diffuse changes in white matter) ( Fig. 1) [15,16]. Perifocal edema associated with a brain metastasis was not included in this assessment. Leukoencephalopathy is by definition diffuse and periventricular, compared to perifocal edema associated with tumors.

Statistical analysis
Median or mean and range were used for continuous variables while frequency and percentages were used for categorical variables. The Kaplan-Meier method was used to assess overall survival (OS) and the incidence of leukoencephalopathy. OS was defined as the time from the first clinical visit until death or the last clinical follow-up visit. The incidence of leukoencephalopathy was defined as the time from the MRI scan associated with the first SRS or WBRT session to the last follow-up MRI scan. OS and the incidence of leukoencephalopathy were compared between the groups (SRS alone versus WBRT plus SRS). Cox regression univariate and multivariate analyses were performed to characterize the correlation potential prognostic indicators (age, sex, KPS, number of metastatic lesions, total tumor volume, extracranial metastasis, baseline leukoencephalopathy, SRS integral dose to the cranium, WBRT, surgical resection, and chemotherapy) and survival or leukoencephalopathy. The Cox proportional hazards model was used to calculate hazard ratios and 95% confidence intervals. Multivariate analysis was used to assess prognostic factors with significance of less than 0.05 in univariate analysis. All statistical analysis was performed using SPSS version 25.0 (SPSS Inc, Taipei, Taiwan).

Clinical presentation
A total of 993 patients were included in the cohort for analysis ( Table 1). The median survival time among the cases of mortality was as follows: SRS alone (13.2 months; range 0.7-151.6 months) and SRS plus WBRT (16.9 months; range 1.6-109.1 months) (p = 0.740). The median followup period was as follows: SRS alone (14.5 months; range 0.1-177.7 months) and SRS plus WBRT (18.4 months; range 0.2-233.2 months). The median imaging followup period was as follows: SRS alone (11.9 months; range 0.1-177.7 months) and SRS plus WBRT (15.8 months; range 0.1-233.2 months) (p = 0.951). The median age was as follows: SRS alone (62.5 years) and SRS plus WBRT (58.1 years) (p = 0.034). The median SRS tumor margin dose was as follows: SRS alone (18 Gy) and SRS plus WBRT (18 Gy). The median total tumor volume was as follows: SRS alone (2.26 mL) and SRS plus WBRT (2.41 mL) (p = 0.189). In the SRS and SRS plus WBRT groups, respectively, the most common origin of BMs was lung cancer (73.2% vs 75.3%) and the most common location of BMs was the frontal lobe (30.9% vs 28.4%), followed by parietal lobe (17.1% vs 15.8%) and cerebellum (15.45% vs 21.8%). We observed no statistical difference between the two groups in terms of median KPS score (p = 0.901).

The incidence of leukoencephalopathy
In Fig. 2, the incidence of leukoencephalopathy is illustrated as a function of time or specific grade. The incidence of leukoencephalopathy in the WBRT plus SRS group increased gradually, as follows: 1st follow-up (18.2%), 2nd follow-up (33.7%), and 3 rd follow-up (50.5%). Patients who underwent WBRT experienced leukoencephalopathy of higher severity with a more abrupt onset. More than 96% of the patients who underwent SRS alone remained free from leukoencephalopathy (grade 0) after a 4-year follow-up.

Prognostic factors for leukoencephalopathy
The prognostic factors for the incidence of leukoencephalopathy are listed in Table 2. WBRT treatment had a significant influence on the likelihood of developing leukoencephalopathy (p < 0.001) ( Table 2 and Fig. 3B). Age (p = 0.025 and p < 0.001) was also associated with an elevated incidence of leukoencephalopathy (Fig. 3C). Finally, a larger total tumor volume was correlated with the incidence of leukoencephalopathy (Fig. 3D), particularly in larger tumors (≧28cm 3 ; p = 0.028 and p < 0.001). Pre-SRS chemotherapy (p = 0.02 and 0.166) and the total number of metastases (p = 0.008 and 0.303) were weakly and significantly correlated in univariate analyses but not significant in multivariate analyses. Other prognostic factors, including sex, extracranial metastasis, and tumor resection, presented no significant correlation with leukoencephalopathy.
Subgroup analysis was used to evaluate the dosimetric effects of WBRT or SRS on leukoencephalopathy. In the SRS alone group, the total number of metastatic lesions, total tumor volume, and integral dose to the skull were not significantly correlated with the incidence of leukoencephalopathy (Supplementary Table 1, Fig. 3F). In the WBRT plus SRS group, a higher total tumor volume was correlated with the incidence of leukoencephalopathy, particularly in larger tumors (≧28cm 3 ; p < 0.001) (Supplementary Table 2, Fig. 3E). Dose plan and integral dose were not correlated with the occurrence of leukoencephalopathy.
Overall, WBRT was the most important factor in the development of leukoencephalopathy, followed by patient age and total tumor volume.
The median survival was as follows: WBRT plus SRS (24.  (41.5%), 4-year (33.0%), and 5-year (28.8%). At any given time, the OS in the SRS alone group was higher than that in the WBRT plus SRS group (p = 0.009) (Fig. 4F). The factor analysis is detailed in Supplementary Tables 1 and 2.

Discussion
In this study, WBRT (in conjunction with SRS) was shown to increase the incidence of leukoencephalopathy compared to SRS alone, while reducing OS rates. These results were in line with those in most previous studies [5,6]. However, the SRS integral dose to the skull presented a non-significant correlation to the development of leukoencephalopathy in both subgroups, which indicated that WBRT is a major factor affecting the onset of leukoencephalopathy. Additionally, two relatively small retrospective studies (n = 103 and 92) reported that SRS was associated with a reduced risk of developing leukoencephalopathy [7,8]. SRS provided focal high radiation energy without inducing changes in white matter. To the best of our knowledge, this is the first large-scale study to demonstrate that SRS is safe in terms of leukoencephalopathy development following radiation therapy for BMs. Concerns related to the side effects of WBRT and radiation-induced neurocognitive dysfunction have prompted several research groups to identify the prognostic factors for radiation-induced leukoencephalopathy, including WBRT dose and fractionation, the number of tumors, tumor size, total tumor volume and location, SRS prescription and KPS Karnofsky Performance Status, SRS stereotactic radiosurgery, WBRT whole-brain radiation therapy *Data are expressed as number of patients or median value, unless otherwise noted #This category excluded patients who survived the entire study. A total of 434 patients were excluded and 559 patients (357 patients with SRS alone and 202 patients with WBRT + SRS) were evaluated for median survival time. The range of median survival time was calculated in the SRS alone and WBRT + SRS groups † All patients in the study underwent at least one SRS procedure. The dose was determined by the integral dose to the brain and prescribed to the highest isodose line encompassing the target. The integral dose was calculated as the product of the mean dose to the outer head contour and the outer head contour volume  integral dose to the brain [17][18][19]. The most important factor in selecting WBRT or SRS has been the numbers of tumors, followed by tumor size, cumulative brain burden, extracranial disease status, and radiotherapeutic dosimetry [20]. However, the conclusions in previous studies have been inconsistent with regard to the cutoff number of tumors in selecting SRS or WBRT more likely due to the number of tumors which is not major factors for evaluation of radiationinduced neurocognitive dysfunction. Thus, most previous studies dealing with the complications of radiosurgery have focused on extracranial disease status, brain tumor burden, the number of tumors, neurocognitive dysfunction, or quality of life [9,21,22]. One recent study identified tumor volume as a prognostic factor in cases where multiple BMs were treated using SRS [22]. Total tumor volume has also been identified as a significant predictor of radiation toxicity; however, it has seldom been considered in studies on the complications of radiosurgery [18,23]. In the current study, total tumor volume was positively correlated with the development of leukoencephalopathy, particularly in cases where the total tumor volume exceeded 28 cm 3 (Table 2). Therefore, the total tumor volume could be regarded as an important prognostic factor to evaluate radiation-induced leukoencephalopathy.
It should be noted that 33 patients in the SRS alone group developed leukoencephalopathy, 15 of whom developed leukoencephalopathy before undergoing SRS treatment (Fig. 2). Although the pathophysiology of leukoencephalopathy was not totally understood [24,25], age was thought to be a prognostic factor related to the development leukoencephalopathy in this study (Table 2 and Fig. 3C). Leukoencephalopathy is not uncommon among cancer patients undergoing chemotherapy (e.g., methotrexate, fluorouracil, or fludarabine) [26]. Leukoencephalopathy secondary to endothelial Fig. 3 Kaplan-Meier estimates indicating the cumulative incidence of leukoencephalopathy for all patients (n = 993) as well as those in the SRS alone subgroup (n = 702) and WBRT plus SRS subgroup (n = 291): A Cumulative incidence of leukoencephalopathy for all patients (n = 993); B Incidence of leukoencephalopathy as a function of treatment regimen (SRS alone versus WBRT plus SRS); C Incidence of leukoencephalopathy as a function of age (≦77 vs. > 77 years old); D Incidence of leukoencephalopathy as a function of total tumor volume (≦14 cm 3 vs. 14-28 cm 3 vs. ≧28 cm 3 ); E Incidence of leukoencephalopathy in the WBRT plus SRS subgroup as a function of total tumor volume (≦14 cm 3 vs. 14-28 cm 3 vs. ≧28 cm 3 ); F Incidence of leukoencephalopathy in the SRS alone subgroup as a function of total tumor volume (< 14 cm 3 vs. ≧14 cm 3 ), where * indicates a difference of statistical significance damage has also been shown to induce the breakdown of the blood-brain barrier, vasculitis, or demyelination [24,25,27]. Above hypotheses had been clinically indicated by evidence demonstrating that impaired glycemic control, and uncontrolled hypertension were important risk factors for endothelial damage [24,25,27]. Thus, it appears that 14-28 cm 3 vs. ≧28 cm 3 ); F OS as a function of treatment regimen (SRS alone versus WBRT plus SRS); G OS as a function of KPS (< 80 vs. ≧80), where * indicates a difference of statistical significance these clinical factors cannot be omitted when we attempt to conclude the occurrence of leukoencephalopathy in cancer patients.

Study limitations
This study was subject to a number of limitations. First, some of the BM patients who passed away early, showed no signs of leukoencephalopathy in follow-up MR images obtained prior to death. This very likely led us to underestimate the overall incidence of leukoencephalopathy and biased the incidence of leukoencephalopathy in the 3 rd or 4 th years (Fig. 2). Second, leukoencephalopathy was commonly identified in MRI scans obtained during routine follow-up, and time-to-event analyses of leukoencephalopathy were relied on when the events are identified. Therefore, we could not warrant whether the occurrence of leukoencephalopathy is at its on-set time or more severe grade. Third, patients in a poor clinical condition may have been treated using WBRT in accordance with the regulations of Taiwan's National Health Insurance Administration, which stipulate cutoff KPS of ≧70 as the criteria for SRS payment. This probably introduced bias in our analysis of OS. Fourth, 4 patients who underwent SRS before WBRT had the latency period over 4 years despite most patients with latency periods less than 1 year in our study, which is attributed to 4 patients who had the interval from SRS to WBRT over 1 year, but none of them developed leukoencephalopathy from the WBRT to leukoencephalopathy initiation over 4 years. Thus, this bias resulted in few cases of the latency periods over 4 years even though most patients with the interval from WBRT to leukoencephalopathy initiation generally less than 4 years, but we were unable to change our initiation point of the latency period because of their low frequency (4 of 993 patients) in the population. Fifth, a better presentation of leukoencephalopathy would be a patient with cognitive dysfunction with typical image feature (so called white matter change). We also recognized that we did not let every terminal patient undergo comprehensive neuropsychiatric tests. In clinical setting, it is also difficult to request all the patients receiving a test only for study. Thus, future studies on the safety of SRS should investigate neurocognitive function, quality of life, and adverse complications after SRS and WBRT.

Conclusions
This study demonstrated that BM patients who receive WBRT face an elevated risk of developing leukoencephalopathy. The incidence of leukoencephalopathy with SRS alone is extremely low compared to those patients who receive combined therapy with whole brain irradiation. Age and total tumor volume were identified as important prognostic factors related to the development of leukoencephalopathy.