Impact of CyberKnife Radiosurgery on Overall Survival and Various Parameters of Patients with 1-3 versus ≥ 4 Brain Metastases

Introduction This study’s objective is to compare the overall survivals (OSs) and various parameters of patients with 1-3 versus ≥ 4 brain metastases post-CyberKnife radiosurgery (CKRS) (Accuray, Sunnyvale, California) alone. Methods Charts of 150 patients, from 2009-2014, treated with only CKRS for brain metastases were reviewed retrospectively for overall survival (OS) and patient, tumor, and imaging characteristics. Parameters included demographics, Eastern Cooperative Oncology Group (ECOG) performance scores, number and control of extracranial disease (ECD) sites, cause of death (COD), histology, tumor volume (TV), and post-CKRS whole brain radiotherapy (WBRT). The imaging characteristics assessed were time of complete response (CR), partial response (PR), stable imaging or local failure (LF), and distal brain failure (DBF). Patients and their data were divided into those with 1-3 (group 1) versus ≥ 4 brain metastases (group 2). For each CR and LF patient, absolute neutrophil count (ANC), absolute lymphocyte count (ALC)), and ANC/ALC ratio (NLR) were obtained, when available, at the time of CKRS. Results Both group 1 and group 2 had a median OS of 13 months. The patient median age for the 115 group 1 patients versus the 35 group 2 patients was 62 versus 56 years. Group 1 had slightly more males and group 2, females. The predominant ECOG score for each group was 1 and the number of ECD sites was one and two, respectively. Uncontrolled ECD occurred in the majority of both group 1 and group 2 patients. The main COD was ECD in both groups. The prevalent tumor histology for groups 1 and 2 was non-small cell lung carcinoma. Median TVs were 1.08 cc versus 1.42 cc for groups 1 and 2, respectively. The majority of patients in both groups did not undergo post-CKRS WBRT. Imaging outcomes were LC (CR, PR, or stable imaging) in 93 (80.9%) and 26 (74.3%) group 1 and 2 patients, of whom 32 (27.8%) and six (17.1%) had CR; 38 (33.0%) and 18 (51.4%), PR and 23 (20.0%) and two (5.7%), stable imaging; LF was the outcome in 22 (19.1%) and nine (25.7%) patients, and DBF occurred in 62 (53.9%) and 21 (60.0%), respectively. Uni- and multivariable analyses showed the independent parameters of a lower ECOG score, a greater number of ECD sites and uncontrolled ECD were significantly associated with greater mortality risk with and without accounting for other covariates. At CKRS, 19 group 1 and 2 CR patients had a mean ANC of 5.88 K/µL and a mean ALC of 1.31 K/µL and 13 (68%) of 19 had NLRs ≤ five, while 11 with LFs had a mean ANC of 5.22 K/µL and a mean ALC of 0.93 K/µL and seven (64%) had NLRs > five. An NLR ≤ five and high ALC was associated with a CR and an NLR > five and a low ALC with an LF. Conclusions Median OS post-CKRS was 13 months for both patients with 1-3 brain metastases and with ≥ 4. This is the only study in the literature to evaluate OS in patients with 1-3 and ≥ 4 brain metastases who were treated with CKRS alone. For groups 1 and 2 patients combined, 119 (79.3%) had LC and 38 (25.3%) had CR. The ANC, ALC, and NLR values are likely predictive of CR and LF outcomes


Results
Both group 1 and group 2 had a median OS of 13 months. The patient median age for the 115 group 1 patients versus the 35 group 2 patients was 62 versus 56 years. Group 1 had slightly more males and group 2, females. The predominant ECOG score for each group was 1 and the number of ECD sites was one and two, respectively. Uncontrolled ECD occurred in the majority of both group 1 and group 2 patients. The main COD was ECD in both groups. The prevalent tumor histology for groups 1 and 2 was non-small cell lung carcinoma. Median TVs were 1.08 cc versus 1.42 cc for groups 1 and 2, respectively. The majority of patients in both groups did not undergo post-CKRS WBRT. Imaging outcomes were LC (CR, PR, or stable imaging) in 93 (80.9%) and 26 (74.3%) group 1 and 2 patients, of whom 32 (27.8%) and six (17.1%) had CR; 38 (33.0%) and 18 (51.4%), PR and 23 (20.0%) and two (5.7%), stable imaging; LF was the outcome in 22 (19.1%) and nine (25.7%) patients, and DBF occurred in 62 (53.9%) and 21 (60.0%), respectively. Uni-and multivariable analyses showed the independent parameters of a lower ECOG score, a greater number of ECD sites and uncontrolled ECD were significantly associated with greater mortality risk with and without accounting for other covariates. At CKRS, 19 group 1 and 2 CR

Introduction
Brain metastases are present at the time of diagnosis of a primary carcinoma in 20-40% of patients [1][2][3][4]. Improved overall survival (OS) is increasing the incidence of brain metastases and is due to novel chemo-and immunotherapies and better imaging techniques, enabling earlier detection of punctate and silent brain lesions [5].
Eighty percent of patients with brain metastases present with 1-3 lesions [6]. Correspondingly, Tsao et al. presented their 2010 survey results analyzing brain metastasis management: 68% of mainly radiation oncologist respondents chose 1-3 brain metastases as the ideal number to treat with stereotactic radiosurgery (SRS) alone. Only 32% treated patients who had ≥ 4 lesions with only SRS [7].
Is the OS of patients who present with ≥ 4 brain metastases worse than that of those with 1-3? Few studies have evaluated the OS post-SRS treatment of patients with newly diagnosed 1-3 or ≥ 4 brain metastases without pre-SRS metastasectomy nor pre-or concurrent-to-SRS whole brain radiotherapy (WBRT) [8][9][10][11][12][13]. Bashir et al. was the only group to compare both arms simultaneously post-Gamma knife radiosurgery (GKR), again, as the sole treatment [14]. The present publication likewise compared both arms' OS, but used CyberKnife radiosurgery (CKRS) (Accuray, Sunnyvale, California) alone.
In the present paper, patients presenting with 1-3 and ≥ 4 brain metastases were evaluated for various parameters known to impact OS. The patient characteristics of age, gender, Eastern Cooperative Oncology Group (ECOG) performance score, number of extracranial disease (ECD) sites, ECD control or non-control at CKRS, alive status, and ECD versus central nervous system (CNS) disease as the cause of death (COD) were evaluated. Tumor characteristics, including histology, total tumor volume (TV) at initial CKRS treatment, and adjunct post-CKRS WBRT were documented. Brain magnetic resonance imaging (MRI) findings and date of brain metastasis local control (LC) (complete response (CR), partial response (PR), and stability) or local failure (LF), all with or without distal brain failure (DBF) were noted. The incidences of leptomeningeal disease (LMD) and radiation necrosis (RN) were also recorded.
In this paper, a method of predicting and facilitating imaging CR is presented. This system utilized CR and, in contrast, LF patients' complete blood counts with differentials as the source of both the absolute neutrophil count (ANC) which was divided by the absolute lymphocyte count (ALC), which is called the neutrophil-to-lymphocyte ratio (NLR). The NLR is a known prognosticator for OS in patients with primary carcinomas. Few papers, however, have utilized the NLR to analyze brain metastasis patients' post-SRS OS and imaging outcomes. Mitsuya et al. evaluated NLRs in pre-craniotomy patients having metastasectomies [15]. An NLR < five portended an OS of 14 months, while the OS was five months with an NLR ≥ five, Shaverdian et al. showed that in pre-SRS patients being treated for brain metastases, on multivariate analysis, a higher peripheral neutrophil percentage predicted poor initial brain MRI, LC, intracranial control of metastases, and OS [16]. A lower percentage of lymphocytes predicted an initial poor MRI and LC. It was not stated that ANCs, ALCs, or NLRs were used in their determinations, however.
As a background for how the ANCs and ALCs may play a role in controlling post-SRS brain metastases, peripheral blood lymphocytes (PBLs) have been found by Schondorff et al. to have similar cell types and cellular characteristics as tumor-infiltrating lymphocytes (TILs) at the site of primary breast and ovarian carcinomas [17]. This finding indicates that PBLs are the likely source of tumor TILs. Szeifert et al. performed histopathological and immunohistochemical evaluations of post-SRS controlled and uncontrolled brain metastases [18]. In the brain metastasis/brain parenchyma interface area and central necrotic region, TILs and tumor-associated neutrophils (TANs) were found, respectively, in controlled brain metastases, and these cells were sparse in these locations in less-well-controlled brain metastases. Thus, the TANs, which have as their likely source, neutrophils (ANCs), and TILs, lymphocytes (ALCs), appear to be involved in brain metastases control post-SRS.
Thus, in the present paper, for the first time and continuing the preliminary work of Schondorff et al., ANC, ALC, and NLR values were determined at the time of CKRS for patients with CR and LF imaging outcomes. This preliminary study was carried out to ascertain if these values were predictive of CR versus LF imaging outcomes.

Study population
Charts of 574 Stanford University Medical Center (SUMC) patients post-CKRS treatment of brain metastases between 2009 and 2014 were reviewed retrospectively after approval by the Institutional Review Board (IRB) of protocol 26173. Excluded were patients who had undergone pre-CKRS metastasectomies or pre-or concurrent-to-CKRS WBRT and those who did not have post-CKRS brain magnetic resonance imaging (MRI) scans.

Data collection
Data for patient characteristics were collected and included each patient's age, gender, ECOG score of 0, 1, or 2, number of ECD sites from 0 to ≥ 4, ECD control or non-control at CKRS, and alive status or COD due to ECD versus CNS disease.
The tumor characteristics of individual histological types, total TV at initial CKRS treatment, and whether the patients had undergone CKRS alone or with post-CKRS WBRT were collected. Prescribed doses (PDs) of CKRS and number of fractions delivered were noted.
Brain MRIs performed immediately pre-and at four-to-six-week intervals throughout post-CKRS treatment determined imaging outcome characteristics of treated brain metastases using neuroradiology reports and target measurements. Measurements were made employing the electronic caliper function of the Centricity picture archiving and communication systems (PACS) (General Electric Healthcare, Milwaukee, WI), which allows a spatial resolution to 0.1 mm. The response assessment in neuro-oncology brain metastases (RANO-BM) method had the following adaptations: cases were included that had ≥ five CKRS target lesions [19]. The sum (of the) longest diameter (SLD) of each CKRS-treated target lesion at the time of CKRS was compared with the SLD at the time of occurrence of CR (disappearance of all CKRS target lesions) without DBF (new brain metastases outside the treated TVs) or at concurrent CR, PR (30% SLD decrease), stable disease (neither PR nor LF), or LF (20% SLD increase) with DBF or final imaging. The time between CKRS and imaging outcome was calculated. Local control (LC) was CR, PR, or stable brain MRI outcomes. The durations from CKRS treatment date to time of occurrence of RN using brain MRI plus histological verification and imagingdetermined-LMD were noted.

Preliminary prognostic marker determinations
Using the complete blood count (CBC) with differential when available at the time of CKRS treatment, the ANC and ALC counts were obtained. The ANC/ALC ratio (NLR) was calculated for each patient who had CR and LF imaging outcomes.

Statistical analyses
Patients with brain metastases treated with CKRS alone were divided into two groups: those who had 1-3 brain metastases targeted at the time of initial CKRS presentation (group 1) versus patients who had ≥ 4 (group 2). Continuous characteristics were summarized as medians with interquartile range (IQR) and categorical characteristics as counts and percentages.
Homogeneity was assessed for patient characteristics (age, sex, ECOG score, number of ECD sites, and COD), tumor characteristics (histology, total tumor volume, post-CKRS WBRT status) and brain MRI responses between groups 1 and 2. For the latter analyses, the Chi-square test (or the Fisher's exact test when the cell count was low) for categorical variables and the Mann-Whitney U test for continuous variables were used. Histological types included non-small cell lung carcinoma (NSCLC), breast, melanoma, renal cell carcinoma (RCC), and "other" carcinomas (bladder, gastric, colorectal, thyroid, ovarian, testicular carcinoma, or tongue and nasal squamous cell carcinoma).
The brain MRI outcomes, including LC and its components, versus LF were also assessed statistically for the number and percentage of each outcome. All patients were followed from CKRS until death or May 1, 2017 (end of the database). Kaplan-Meier estimates were computed by group (1-3 versus ≥ 4 brain metastases), and the OSs were compared using the logrank test. The association between the brain metastasis group and mortality was also assessed using uniand multivariable Cox proportional hazards models, which adjusted for various patient and tumor characteristics. All statistical tests were performed at the 0.05 significance level, and analyses were implemented using the R-3.3 software [20].

Uni-and multivariable analyses
Using uni-and multivariable analyses, a poor ECOG score, greater ECD burden, and uncontrolled ECD status were significantly associated with a lower OS, with and without accounting for other covariates. Uni-and multivariable analyses confirmed that men were at significantly greater risk for a lower OS when gender was assessed alone, but not so in adjusted analyses. Also, when using a multivariate analysis, group status (1 versus 2) was not statistically significantly associated with OS after adjusting for various confounding factors (

Discussion
The most common number of brain metastases at presentation is 1-3 [6]. A recent survey by Tsao et al., documenting predominantly radiation oncologists' responses, showed that the maximum number of brain metastases they would treat with SRS was 3 [7]. Few publications have compared OSs after SRS treatment without pre-SRS metastasectomies nor pre-SRS or concurrent-with-SRS WBRT of patients presenting with 1-3 brain metastases versus those presenting with ≥ 4 to refute the latter management scheme.

Literature review
Overall survival: A PubMed literature review of publications that assessed OS after SRS without pre-SRS metastasectomies nor pre-SRS or concurrent WBRT in patients presenting with 1-3 versus ≥ 4 brain metastases was performed. The terms stereotactic, Gamma Knife (Elekta, Stockholm, Sweden), CyberKnife, linear accelerator (LINAC), radiosurgery, brain metastases, outcomes, survival, mortality, 1-3, and four and more were used. The search was limited to English language articles only, but not by date, age group, or type of publication. This review found that of seven investigators (after excluding the present paper), five evaluated OS for solely the 1-3 brain metastases group [8][9][10][11][12]. Conversely, only one group studied OS only in patients presenting with ≥ 4 brain metastases (   Thus, the five groups who evaluated the OS of patients with 1-3 brain metastases included Schuttrumpf et al., who used a LINAC system alone to treat 189 patients with 1-3 brain metastases and the median OS was 232 days (7.6 months) [12]. Brown et al. documented 111 patients with ECOG scores of 0-2 and ≤ three solely-SRS-treated brain metastases at presentation, who had a median OS of 10.4 months [9]. Lutterbach et al. presented the results of their review of 101 patients, also with only patients having ≤ three brain metastases treated with an Elekta alone. The latter group of patients had a median OS of 7.6 months [11]. Rades et al. stated that, in general, the outcome for patients with 1-3 brain metastases "appears to be better than that for patients with > 3 lesions" [8]. This group, however, analyzed the median OS of 95 patients, but also only those with 1-3 brain metastases, who had recursive partitioning analysis (RPA) scores of 1-2. They used linear accelerator-based SRS alone or Gamma knife radiosurgery (GKR) alone to attain a median OS of 13 months. El Ganterey et al. utilized a LINAC device alone to treat 18 patients with 1-3 brain metastases, who had a median OS of eight months [10].
The publication by Ojerholm et al. addressed OS in 38 patients who had only the higher number of brain metastases, i.e. ≥ 4 [13]. Their patients were RPA Class 2-3 and were treated with GKR alone. These patients attained an OS of 6.7 months.
Bashir et al. were the only authors, besides those of the present paper, to determine the median OS for patients with 1-3 brain metastases (eight months) and included a separate OS analysis of those with ≥ 4 (7.5 months) [14]. These authors, however, used GKR, again alone, rather than CKRS, as in the present study.
The present paper addresses the paucity of publications that simultaneously evaluate the median OSs of patients with both 1-3 and ≥ 4 lesions, who were treated with CKRS, without prior craniotomies for brain metastases resection or pre-or concurrent with CKRS WBRT. The mean OS for the five studies that evaluated the 1-3 arm was 9.3 months. The median OS was 6.7 months for patients with ≥ 4 brain metastases per a single study by Ojerholm et al., who evaluated this group. The difference between the mean OS for each of the two groups above of 9.3 months versus 6.7 may be due to the relatively small number of 38 patients in the ≥ 4 brain metastasis group versus the total of 514 patients who had 1-3 brain metastases at presentation in the other five studies.
For this paper's 115 patients in group 1, the median OS was 13 months, while 35 patients in group 2 also had a median OS of 13 months ( Table 7, Figure 1).

Statistical analyses
After adjusting for various confounding factors using a multivariable analysis, statistical analysis of the data from the present study showed that the number of brain metastases at the time of presentation is not significantly associated with OS, while the independent parameters, however, of uncontrolled ECD, greater number of ECD sites, and a lower ECOG score were, with and without accounting for other covariates. Men were at significantly great risk when gender was assessed alone, but, again, not in the adjusted analyses. The latter gender greater risk may be impacted in our study by the fact that the majority of patients had NSCLC as their histological diagnosis. Ferguson et al. documented a male preponderance of 478 (62%) versus 294 (38%) women in their study of gender-associated differences in patients with lung cancer [22]. Women with NSCLC had a better OS of 8.5 months versus 5.4 for men. The present paper's group 1 NSCLC patients had a preponderance of 39 (62%) men versus 24 (38%) females, so this likely negatively impacted the median OS for this group 1 since women have better OSs with this disease. For the 25 NSCLC patients with ≥ 4 brain metastases in the present study, however, there was a predominance of 14 women (56%) versus 11 men (44%), which may have slightly positively skewed the OS of group 2 via the NSCLC category, though the impact of the latter was less due to the lower number of patients in group 2. Of 115 patients in the present study with 1-3 brain metastases at presentation, 93 (80.9%) had overall LC and of 35 patients who had ≥ 4 brain metastases, 26 (74.3%) achieved this. Our LCs for both groups may be even higher since some of the LFs likely were instances of pseudoprogression, which were categorized as LFs, rather than LCs.

Complications
Of 101 patients with 1-3 brain metastases treated with a LINAC system alone, Lutterbach et al. noted one incidence (1%) of "likely RN," though this was not documented histologically due to the patient's expiration [11]. Of 38 solely GKR-treated patients evaluated by Ojerholm et al., only one RN (3%) case occurred also [13]. In the present study, five (4.3%) instances of histologically documented RN were noted in group 1 patients only; our patient number of 150 was much higher than the numbers of the latter two studies. Szeifert et al. have shown that TANs and TILs are both present in well-controlled SRS-treated brain metastases; these cells have pro-and anti-tumor effects, respectively, per Coffelt et al. [18,23]. These latter authors reported that systemic neutrophil depletion in mice with mammary breast cancers lead to a reduction in lung metastases. They also demonstrated in these mice that the mammary tumors produce interleukin (IL)1β, which induces γδ T cells to produce IL17. The increased systemic IL17 level upregulates granulocyte-colony stimulating factor (G-CSF), which drives the systemic expansion and alteration of neutrophil phenotypes. These altered neutrophils produce inducible nitric oxide synthase (iNOS), which suppresses an anti-tumor cluster of differentiation 8 (CD8) T cell activity (CD8+ T cells). This results in subsequent increased chances of metastasis formation in distant organs. Thus, this cascade of events likely occurs in the post-SRS brain metastases. This cascade may actually contribute to the control of post-SRS brain metastases by TILs. The number of TILs may be reduced by the TANs so as not to cause overstimulation of the TILs, which would cause pseudo-progression.

Increasing CR outcomes using ALC and NLR values
Stereotactic radiosurgery also stimulates TILs per Szeifert et al., and this may occur by SRS blood-brain barrier damage. This damage allows lymphoid infiltration into the perivascular space of the brain surrounding the tumor. They noted that SRS causes brain metastasis cells to act as an increased activation stimulus for systemic T cells, resulting in the abscopal effect [18].
ALC mean values in CR versus LF patients: The CR patients' CKRS ALC mean was 1.31 K/µL, while the LF patients' mean value was 0.93 K/µL The CR patients thus likely have a higher reservoir of lymphocytes than the LF patients for brain metastases control by TILs post-SRS. This postulation would correlate with Szeifert et al.'s findings of more TILs surrounding controlled brain metastases and the converse in the uncontrolled brain metastases described above.
NLR mean values in CR versus LF patients: A preliminary analysis of NLRs ≤ 5 and > 5 at the time of CKRS treatment for CR and LF imaging outcome patients was carried out in the present paper as one of the first in the literature. Of 19 groups 1 and 2 CR imaging outcome patients, the majority or 13 (68%) had NLRs ≤ 5. Of 11 groups 1 and 2 LF imaging outcome patients, the majority or seven (64%) had NLRs > five. Thus, NLRs ≤ five were associated with a higher chance of a CR imaging outcome. This reflects the higher mean ANC numerator value of 5.88 K/µL in the CR patients versus 5.22 K/µL in the LF patients and the higher ALC denominator value of 1.31 K/µL for the CR patients versus 0.93 K/µL for the LF patients, obtained from individual NLR calculations for each patient.
The mean steroid administration for the interval between CKRS and CR outcome was 64 mg, while for the LF patients, 23 mg. Cook et al. evaluated peripheral blood mononuclear cells after decadron administration and found that the regulatory T cells underwent significant increase and proliferation [24]. Tanaka et al. stated that the regulatory T cells suppress the anti-tumor immune response [25]. Alternatively, glucocorticoids also enhance the activity of macrophages and induce tolerogenic dendritic cells, exerting a potent anti-inflammatory effect [26][27]. The latter is possibly the reason that the CR patients who received larger amounts of steroids than did the LF patients did so well regarding their imaging outcomes since Szeifert et al. showed that macrophages were found with plasmocytes at SRS-treated brain metastases, which were controlled [18].
Techniques to increase CR outcomes: The number of patients with CR could be increased if the patients had monitoring of ANC, ALC, and NLR values during the pre-and post-CKRS time intervals. These values would be used to determine and possibly to adjust the chemo-and immunotherapies and other agents, such as steroids, which patients undergoing CKRS for brain metastases receive, the latter with possibly deleterious effects on ALCs and possibly ANCs. The steroid beneficial effects on the patients' ANC, ALC, and NLR levels needs further study with a larger number of CKRS-treated brain metastasis patients. On the other hand, the patients received adjunctive therapies, such as Gemzar and other agents ( Table 6), all of which may lower the ALCs, but, in turn, also have immuno-stimulatory properties [28]. Thus, the ANC, ALC, and NLR evaluation appears to be promising and needs to be studied further in more patients with CKRS-treated brain metastases to gain more insight into how the CR outcomes of CKRS can be influenced and increased.